AI For Advertisers: How Data Analytics Can Change The Maths Of Advertising?

All Images Credit: Freepik

The task of understanding a customer’s journey and designing your marketing strategy accordingly can be difficult in this data-driven world. Today, the customer expresses their needs in myriad forms of requests.

Consumers express their needs and want attitudes, and values in various forms through search, comments, blogs, Tweets, “likes,” videos, and conversations and access such data across many channels like web, mobile, and face to face. Volume, variety, velocity and veracity of the data accumulated through these customer interactions are huge.

BigData and data analytics can be leveraged to understand several phases of the customer journey. There are risks involved in using Artificial Intelligence for the marketing data analysis of data breach and even manipulation. But, AI do have brighter prospects when it comes to marketing and advertiser applications.

As the CEO of a technology firm Chop Dawg and marketer, Joshua Davidson puts it, “AI-powered apps are going to be the future for us, and there are several industries that are ripe for this.” The mobile-first strategy of many enterprises has powered the use of AI for digital marketing and developing technologies and innovations to power industries with intelligent systems.

How AI and Machine learning are affecting customer journeys?

Any consumer journey begins with the recognition of a problem and then stages like initial consideration, active evaluation, purchase, and postpurchase come through up till the consumer journey is over. The need for identifying the purchasing and need patterns of the consumers and finding the buyer personas to strategize the marketing for them.

Need and Want Recognition:

Identifying a need is quite difficult as it is the most initial level of a consumer’s journey and it is more on the category level than at a brand level. Marketers and advertisers are relying on techniques like market research, web analytics, and data mining to build consumer profiles and buyer’s persona for understanding the needs and influencing the purchase of products. AI can help identify these wants and needs in real-time as the consumers usually express their needs and wants online and help build profiles more quickly.

AI technologies offered by several firms help in consumer profiling. Firms like Microsoft offers Azure that crunches billions of data points in seconds to determine the needs of consumers. It then personalizes web content on specific platforms in real-time to align with those status-updates. Consumer digital footprints are evolving through social media status updates, purchasing behavior, online comments and posts. Ai tends to update these profiles continuously through machine learning techniques.

Initial Consideration:

A key objective of advertising is to insert a brand into the consideration set of the consumers when they are looking for deliberate offerings. Advertising includes increasing the visibility of brands and emphasize on the key reasons for consideration. Advertisers currently use search optimization, paid search advertisements, organic search, or advertisement retargeting for finding the consideration and increase the probability of consumer consideration.

AI can leverage machine learning and data analytics to help with search, identify and rank functions of consumer consideration that can match the real-time considerations at any specific time. Take an example of Google Adwords, it analyzes the consumer data and helps advertisers make clearer distinctions between qualified and unqualified leads for better targeting.

Google uses AI to analyze the search-query data by considering, not only the keywords but also context words and phrases, consumer activity data and other BigData. Then, Google identifies valuable subsets of consumers and more accurate targeting.

Active Evaluation: 

When consumers narrow it down to a few choices of brands, advertisers need to insert trust and value among the consumers for brands. A common technique is to identify the higher purchase consumers and persuade them through persuasive content and advertisement. AI can support these tasks using some techniques:

Predictive Lead Scoring: Predictive lead scoring by leveraging machine learning techniques of predictive analytics to allow marketers to make accurate predictions related to the intent of purchase for consumers. A machine learning algorithm runs through a database of existing consumer data, then recognize trends and patterns and after processing the external data on consumer activities and interests, creates robust consumer profiles for advertisers.

Natural Language Generation: By leveraging the image, speech recognition and natural language generation, machine learning enables marketers to curate content while learning from the consumer behavior in real-time scenarios and adjusts the content according to the profiles on the fly.

Emotion AI: Marketers use emotion AI to understand consumer sentiment and feel about the brand in general. By tapping into the reviews, blogs or videos they understand the mood of customers. Marketers also use emotion AI to pretest advertisements before its release. The famous example of Kelloggs, which used emotion AI to help devise an advertising campaign for their cereal, eliminating the advertisement executions whenever the consumer engagement dropped.


As the consumers decide which brands to choose and what it’s worth, advertising aims to move them out of the decision process and push for the purchase by reinforcing the value of the brand compared with its competition.

Advertisers can insert such value by emphasizing convenience and information about where to buy the product, how to buy the product and reassuring the value through warranties and guarantees. Many marketers also emphasize on rapid return policies and purchase incentives.

AI can completely change the purchase process through dynamic pricing, which encompasses real-time price adjustments on the basis of information such as demand and other consumer-behavior variables, seasonality, and competitor activities.


Aftersales services can be improved through intelligent systems using AI technologies and machine learning techniques. Marketers and advertisers can hire dedicated developers to design intelligent virtual agents or chatbots that can reinforce the value and performance of a brand among consumers.

Marketers can leverage an intelligent technique known as Propensity modeling to identify the most valuable customers on the basis of lifetime value, likelihood of reengagement, propensity to churn, and other key performance measures of interest. Then advertisers can personalize their communication with these customers on the basis of these data.


AI has shifted the focus of advertisers and marketers towards the customer-first strategies and enhanced the heuristics of customer engagement. Machine learning and IoT(Internet of Things) has already changed the way customer interact with the brands and this transition has come at a time when advertisers and marketers are looking for new ways to tap into the customer mindset and buyer’s persona.

All Images Credit: Freepik

Best machine learning algorithms you should know

Machine learning is a key technology tool businesses use to build tools that enhance their operations. To do that, they take advantage of machine learning algorithms that come in different shapes and sizes, servicing different purposes and working on different data sets. Choosing the right algorithm for the job is what makes machine learning and deep learning projects successful. That’s why being aware of all the different types of machine learning algorithms is so important – that’s how you get better results and build more advanced solutions.

Here’s an overview of the best machine learning algorithms you should know before starting your project.

What is meant by machine learning algorithms?

First things first, what is machine learning and how do algorithms fit into the picture? A machine learning (ML) algorithm is a process or set of procedures that allow a model to adapt to the data with a specific objective set as the goal.

An ML algorithm specifies how the data is transformed from the input to output, helping the model to learn the appropriate mapping from input to output. That model specifies the mapping functions and holds the parameters in place, while the machine learning algorithm updates the parameters to help the model match its goal.

What are the algorithms used in machine learning?

Algorithms can model problems in many different ways. The easiest way to differentiate between different ML algorithms is by comparing them by learning styles that they can adapt. Generally, machine learning algorithms can adapt to several learning styles that help to solve different problems.

Here are four learning styles in machine learning you need to know:

1 Supervised learning

In supervised learning, the input data serves as training data and comes with a known label or result – for example, the price at a time or spam/not-spam.

In this variant, the training process is critical for preparing a model that makes predictions and then is corrected when the predictions are wrong. The training process continues until the model achieves the appropriate level of accuracy. Classification and regression are examples of problems for this learning type.


2 Unsupervised learning

In unsupervised learning, input data isn’t labeled and doesn’t come with a known result. Data scientists prepare models by deducing the structures in the input data to extract general rules or reduce redundancy through mathematical processes. Unsupervised learning addresses problems such as association rule learning, dimensionality reduction, and clustering.

3 Semi-supervised learning

In this learning style, the input data is a mixture of labeled and unlabeled examples. The prediction problem is known, but the model needs to learn the structures for organizing data and making predictions on its own. This learning style is used to address problems such as regression and classification.

4 Reinforcement learning

One of three basic machine learning paradigms together with supervised learning and unsupervised learning, reinforcement learning (RL) is an area of machine learning that focuses on the ways in which software agents should take actions to maximize a specified notion of cumulative reward in a given environment.

The best machine learning algorithms you should know

1 Linear Regression

Linear regression is an algorithm that correlates between two variables in the data set, examining the input and output sets to show a relationship between them. For example, the algorithm can show how changing one of the input variables affects the other variable. The relationship is represented by plotting a line on the graph.

Linear regression is one of the most popular algorithms in machine learning because it’s transparent and requires no tuning to work. Practical applications of this algorithm are risk assessment or sales forecasting solutions.

2 Logistic regression

Logistic regression is a type of constrained Linear Regression with a non-linearity application after you apply weights. Note that this algorithm is used for classification, not regression. The algorithm restricts the outputs close to +/- classes (and 1 and 0 in the case of sigmoid) and can be trained with Gradient Descent or L-BFGS.

Logistic regression is used in Natural Language Processing (NLP) applications, where it often appears under the name of Maximum Entropy Classifier.

3 Principal component analysis (PCA or LDA)

Principal component analysis is an unsupervised method that helps data scientists to understand better the global properties of a data set that consists of vectors. It analyzes the covariance matrix of data points to learn which dimensions/data points have high variance among themselves and low covariance with others. The algorithm helps data scientists to get data points with reduced dimensions.

4 K-means clustering

K- means clustering is a type of unsupervised clustering algorithm that sorts data sets through defined clusters. It offers results in the form of groups based on internal patterns.

For example, you can use a K-means algorithm for sorting web results for the word “cat,” and it will show all the results in the form of groups. The main advantage of this algorithm is its accuracy as it provides data groupings faster than other algorithms.


5 Decision trees

A decision tree is made of various branches that represent the outcome of many decisions. This algorithm collects and graphs data in multiple branches to predict response variables on the basis of past decisions. It comes in handy for mapping our decisions and presents results visually to communicate findings easily.

Decision trees work best for smaller data sets and relatively low-stake decisions – otherwise, the long-tail visuals can be hard to decipher. The key advantage of this algorithm is that it allows showing multiple outcomes and tests without having to involve data scientists – it’s easy to use.

6 Random forests

A random forest consists of a great number of individual decision trees where they all operate as an ensemble. An individual tree in the random forest generates a class prediction – the class which receives the highest number of votes becomes the model’s prediction. Having many relatively uncorrelated models (trees) operating as a committee easily outperforms individual constituent models.

The low correlation between these models is the strength of this approach because it allows producing ensemble predictions that are far more accurate than individual predictions. Note that decisions trees protect each other from individual errors. While some trees may generate false predictions, others will generate the right ones – as a group; they will be able to move in the right direction.

7 Support Vector Machine

Support Vector Machines (SVMs) are linear models similar to linear or logistic regression we’ve discussed earlier. However, there’s one difference – they have a different margin-based loss function, which can be optimized by using methods such as L-BFGS or SGD. SVMs internally analyze data sets into classes, which is helpful for future classifications.

The main idea behind SVM is separating data into classes and maximizing the margins of entering future data into classes. This type of algorithm works best for training data. However, it can also serve as a tool for processing nonlinear data. The financial sector makes use of Support Vector Machines thanks to its accuracy in classifying both current and future data sets.

8 Apriori

The Apriori algorithm is used a lot in market analysis. It’s based on the principle of Apriori and checks for positive and negative correlations between products after analyzing values in data sets.

For example, if two values often correlate in a data set, the algorithm will conclude that A will often lead to B, referring to the information in data sets. For example, if customers often buy product A and product B together, this relation will hold a high percentage and help companies like Google or Amazon to predict product searches and purchases.

9 Naive Bayes Classifier

This handy classification technique is based on Bayes’ Theorem, which assumes independence among predictors. The algorithm will assume that the presence of a specific feature in a class is not related to the presence of any other feature in the same class.

For example, a fruit may be considered a banana if it’s yellow, curved, and about 15 cm long. These features depend on each other, and on the existence of hooter features, they all independently contribute to the probability that this fruit is a banana. That’s why the algorithm bears the name “Naive.”

The algorithm offers a model that is easy to build and helpful in handling very large data sets. It can outperform the most sophisticated classification methods.

10 K-Nearest Neighbors (KNN)

This is one of the simplest algorithm types used in machine learning for classification and regression. KNN algorithms classify new data points on the basis of similarity measures, such as the distance function. They perform classification by using a majority vote of the data points’ neighbors. They then assign data to the class, which has the nearest neighbors. Together with increasing the number of nearest neighbors (the value of k), the accuracy may increase as well.

11 Ordinary Least Squares Regression (OLSR)

Ordinary Least Squares Regression (OLSR) is a generalized linear modeling technique data scientists use for estimating unknown parameters that are part of a linear regression model. OLSR describes the relationship between a dependent variable and one or more of its independent variables.

The algorithm is applied in diverse fields such as economics, finance, medicine, and social sciences. Companies use it in machine learning and predictive analytics to dynamically predict specific outcomes on the basis of variables that change dynamically.

We hope that this machine learning algorithms list helps you pick the right tools of the trade for your next machine learning project. If you’d like to learn more about Machine Learning, Data Science and Web Development, visit the Sunscrapers company blog.

Interview: Data Science im Einzelhandel

Interview mit Dr. Andreas Warntjen über den Weg zum daten-getriebenen Unternehmen – Data Science im Einzelhandel

Zur Einführung der Person:

Dr. Andreas Warntjen arbeitet seit Juli 2016 bei der Thalia Bücher GmbH, aktuell als Senior Manager Advanced and Predictive Analytics. Davor hat Herr Dr. Warntjen viele Jahre als Sozialwissenschaftler an ausländischen Universitäten geforscht. Er hat selbst langjährige Erfahrung in der statistischen Datenanalyse mit Stata, SPSS und R und arbeitet im Moment mit der in-memory Datenbank SAP HANA sowie Python und SAP’s Automated Predictive Library (APL).

Data Science Blog: Herr Dr. Warntjen, welche Bedeutung hat die Data Science für Sie und Ihren Bereich bei Thalia? Und wie ordnen Sie die verwandten Begriffe wie Predictive Analytics und Advanced Analytics im Kontext der geschäftlichen Entscheidungsfindung ein?

Data Science spielt bei Thalia in unterschiedlichsten Bereichen eine zunehmend größer werdende Rolle. Neben den klassischen Themen wie Betrugserkennung und Absatzprognosen ist für Thalia als Buchhändler Text Mining von zentraler Bedeutung. Das größte Potential liegt aus meiner Sicht darin, besser auf die Wünsche unserer  Kunden eingehen zu können.

Bei Thalia werden in schneller Taktung Innovationen eingeführt. Sei es die Filialabholung, bei der online bestellte Bücher innerhalb von 2 Stunden in einer Buchhandlung abgeholt werden können. Oder das Beratungs- und Bezahl-Tablet für die Mitarbeiter vor Ort. Oder Innovationen im Webshop. Bei der Beurteilung, ob diese Neuerungen tatsächlich Kundenwünsche effektiv und effizient erfüllen, kann Advanced Analytics helfen. Im Gegensatz zur klassischen Business Intelligence – die weiterhin eine wichtige Rolle bei der Entscheidungsfindung im Unternehmen spielen wird – berücksichtigt Advanced Analytics stärker die Vielfalt des Kundenverhaltens und der unterschiedlichen Situationen in den Filialen. Verfahren wie etwa multivariate Regressionsanalyse, Entscheidungsbäume und statistische Hypothesentest können die in Unternehmen etablierte Analyse von deskriptiven Statistiken – etwa der Vergleich von Umsatzzahlen zwischen Pilot- und Vergleichsfilialen mit Pivot-Tabellen – ergänzen.

Predictive Analytics kann helfen verschiedenste Geschäftsprozesse individuell für Kunden zu gestalten. Generell können auf Grundlage von automatischen, in Echtzeit erstellten Vorhersagen Prozesse im Unternehmen optimiert werden. Außerdem kann Predictive Analytics Mitarbeiter bei wiederkehrenden Tätigkeiten unterstützen, beispielsweise in der Disposition.

Data Science Blog: Welche Fähigkeiten benötigen gute Data Scientists denn wirklich zur Geschäftsoptimierung? Wie wichtig ist das Domänenwissen?

Die wichtigsten Eigenschaften eines Data Scientist sind große Neugierde, eine sehr analytische Denkweise und eine exzellente Kommunikationsfähigkeit. Um mit Data Science erfolgreich Geschäftsprozesse zu optimieren, benötigt man ein breites Wissensspektrum: vom Geschäftsprozess über das IT-Datenmodell und das Know-how zur Entwicklung von Vorhersagemodellen bis hin zur Prozessintegration. Das ist nur im Team machbar. Domänenwissen spielt dabei eine wichtige Rolle, weshalb es für den Data Scientist essentiell ist sich mit den Prozessverantwortlichen und Business Analysten auszutauschen.

Data Science Blog: Sie bearbeiten Anwendungsfälle für den Handel. Können sich Branchen die Anwendungsfälle gegenseitig abschauen oder sollte jede Branche auf sich selbst fokussiert bleiben?

Es gibt sowohl Anwendungsfälle, die für den Einzelhandel und andere Branchen gleichermaßen relevant sind, als auch Themen, die für Thalia als Buchhändler besonders wichtig sind.

Die Individualisierung im eCommerce ist ein branchenübergreifendes Thema. Analytisches CRM, etwa das zielsichere Ausspielen von Kampagnen oder eine passgenaue Kundensegmentierung, ist für eine Versicherung oder Bank genauso wichtig wie für den Baumarkt oder den Buchhändler. Die Warenkorbanalyse mit statistischen Algorithmen ist ein klassisches Data Mining-Thema, das für den Einzelhandel generell interessant ist.

Natürlich muss man sich vorab über die Besonderheiten des jeweiligen Geschäftsumfeldes Gedanken machen, aber prinzipiell kann man von Unternehmen oder Branchen lernen, die Advanced und Predictive Analytics schon seit Jahren oder Jahrzehnten nutzen. Die passende IT-Infrastruktur und das entsprechende Interesse vom Fachbereich vorausgesetzt, eignen sich diese Anwendungsfälle damit besonders für den Einstieg in Advanced und Predictive Analytics – auch für Mittelständler.

Das Kerngeschäft des Buchhändlers  Thalia ist es, Kunden mit für sie interessanten Geschichten zusammen zu bringen. Die Geschichten selber bestehen aus Text. Die Produktbeschreibungen („Klappentexte“) und -besprechungen liegen in Textform vor. Und Kundenfeedback – sei es auf oder in sozialen Medien – erreicht uns als Text. Erkenntnisse aus Texten abzuleiten (Text Mining) ist deshalb für Thalia wichtiger als für andere Einzelhändler.

Data Science Blog: Welche Algorithmen und Tools verwenden Sie für Ihre Anwendungsfälle? Womit machen Sie eher gute, womit eher schlechte Erfahrungen?

Die Palette bei Thalia reicht von A wie Automated Machine Learning bis Z wie Zeitreihenanalyse. Ich selber arbeite aktuell mit verschiedenen Klassifikationsalgorithmen (z.B., regularisierte logistische Regression,  Random Forest, XGB, Naive Bayes, SAP’s Automated Predictive Library). Im Bereich Text Mining beschäftigen wir uns im Moment unter anderem mit Topic Models und Word2Vec.

Sowohl Algorithmus als auch die Software muss zum Verwendungszweck passen. Bei der Auswahl des Algorithmus gibt es häufig einen Trade-off zwischen Interpretierbarkeit und Prognosegüte. Das muss zusammen mit der Fachabteilung je nach Anwendungsfall abgewogen werden.

Mit flexibler Open Source-Software wie etwa R oder Python lassen sich schnell Proof-of-Concept-Projekte verwirklichen. Für die Integration in bestehende Prozesse sind manchmal kommerzielle Software-Lösungen besser.

Data Science Blog: Soviel zum kurz- und mittelfristigen Start in die Datennutzung. Wie sieht es für die langfristige Verankerung von Advanced/Predictive Analytics im Unternehmen aus? Was muss hier im Rahmen der IT-Infrastruktur bedacht und verankert werden?

Ohne Daten keine Datenanalyse. Je flexibler man auf unterschiedliche Daten im Unternehmen zugreifen kann, desto höher die Innovationsgeschwindigkeit durch Advanced/Predictive Analytics. „Datensilos“ abzubauen bzw. zu vermeiden ist also ein sehr wichtiges Thema. Hohe Datenqualität und die umfassende Dokumentation von Daten sind auch essentiell. Das gilt natürlich nicht nur für Advanced und Predictive Analytics sondern auch für Business Intelligence.

Die langfristige Verankerung von Advanced und Predictive Analytics im Unternehmen verlangt den Aufbau und die kontinuierliche Weiterentwicklung von Infrastruktur in Form von Hardware, Software, Kompetenzen und Wissen, sowie Organisationsformen und Prozessen. Wertschöpfung durch Advanced bzw. Predictive Analytics erfordert das konstruktive Zusammenspiel von Domänenexpertise aus der Fachabteilung, Wissen über Datenstrukturen und -modellen  aus der IT-Abteilung bzw. BI/BW-Systemen und tiefem statistischem Know-how. Nur durch die Zusammenarbeit verschiedener Unternehmensbereiche entstehen Erfolge für das gesamte Unternehmen.

Data Science Blog: Auch organisatorisch sollte langfristig sicherlich einiges bedacht werden. Wann sollten Projekte in den jeweiligen Fachbereichen direkt umgesetzt werden? Wann vielleicht besser in einer zentralen Daten-Abteilung?

Das hängt von einer Reihe von Faktoren ab. Bei hochgradig spezialisiertem Know-how, von dem unterschiedliche Fachbereiche profitieren können, kann es Synergie-Effekte geben, wenn dies zentral organisiert ist. Eine zentrale Einheit kann vielleicht auch Innovationen breiter in ein Unternehmen tragen. Wenn bestimmte Anwendungsszenarien von Advanced/Predictive Analytics für eine Fachabteilung hingegen eine zentrale Rolle spielen oder sie sich ein einem sehr schnelllebigen Umfeld bewegt, dann wäre eine fachliche und organisatorische Verankerung im Fachbereich wichtig.

Visual Question Answering with Keras – Part 2: Making Computers Intelligent to answer from images

Making Computers Intelligent to answer from images

This is my second blog on Visual Question Answering, in the last blog, I have introduced to VQA, available datasets and some of the real-life applications of VQA. If you have not gone through then I would highly recommend you to go through it. Click here for more details about it.

In this blog post, I will walk through the implementation of VQA in Keras.

You can download the dataset from here: All my experiments were performed with VQA v2 and I have used a very tiny subset of entire dataset i.e all samples for training and testing from the validation set.

Table of contents:

  1. Preprocessing Data
  2. Process overview for VQA
  3. Data Preprocessing – Images
  4. Data Preprocessing through the spaCy library- Questions
  5. Model Architecture
  6. Defining model parameters
  7. Evaluating the model
  8. Final Thought
  9. References

NOTE: The purpose of this blog is not to get the state-of-art performance on VQA. But the idea is to get familiar with the concept. All my experiments were performed with the validation set only.

Full code on my Github here.

1. Preprocessing Data:

If you have downloaded the dataset then the question and answers (called as annotations) are in JSON format. I have provided the code to extract the questions, annotations and other useful information in my Github repository. All extracted information is stored in .txt file format. After executing code the preprocessing directory will have the following structure.

All text files will be used for training.


2. Process overview for VQA:

As we have discussed in previous post visual question answering is broken down into 2 broad-spectrum i.e. vision and text.  I will represent the Neural Network approach to this problem using the Convolutional Neural Network (for image data) and Recurrent Neural Network(for text data). 

If you are not familiar with RNN (more precisely LSTM) then I would highly recommend you to go through Colah’s blog and Andrej Karpathy blog. The concepts discussed in this blogs are extensively used in my post.

The main idea is to get features for images from CNN and features for the text from RNN and finally combine them to generate the answer by passing them through some fully connected layers. The below figure shows the same idea.


I have used VGG-16 to extract the features from the image and LSTM layers to extract the features from questions and combining them to get the answer.

3. Data Preprocessing – Images:

Images are nothing but one of the input to our model. But as you already may know that before feeding images to the model we need to convert into the fixed-size vector.

So we need to convert every image into a fixed-size vector then it can be fed to the neural network. For this, we will use the VGG-16 pretrained model. VGG-16 model architecture is trained on millions on the Imagenet dataset to classify the image into one of 1000 classes. Here our task is not to classify the image but to get the bottleneck features from the second last layer.

Hence after removing the softmax layer, we get a 4096-dimensional vector representation (bottleneck features) for each image.

Image Source:


For the VQA dataset, the images are from the COCO dataset and each image has unique id associated with it. All these images are passed through the VGG-16 architecture and their vector representation is stored in the “.mat” file along with id. So in actual, we need not have to implement VGG-16 architecture instead we just do look up into file with the id of the image at hand and we will get a 4096-dimensional vector representation for the image.

4. Data Preprocessing through the spaCy library- Questions:

spaCy is a free, open-source library for advanced Natural Language Processing (NLP) in Python. As we have converted images into a fixed 4096-dimensional vector we also need to convert questions into a fixed-size vector representation. For installing spaCy click here

You might know that for training word embeddings in Keras we have a layer called an Embedding layer which takes a word and embeds it into a higher dimensional vector representation. But by using the spaCy library we do not have to train the get the vector representation in higher dimensions.


This model is actually trained on billions of tokens of the large corpus. So we just need to call the vector method of spaCy class and will get vector representation for word.

After fitting, the vector method on tokens of each question will get the 300-dimensional fixed representation for each word.

5. Model Architecture:

In our problem the input consists of two parts i.e an image vector, and a question, we cannot use the Sequential API of the Keras library. For this reason, we use the Functional API which allows us to create multiple models and finally merge models.

The below picture shows the high-level architecture idea of submodules of neural network.

After concatenating the 2 different models the summary will look like the following.

The below plot helps us to visualize neural network architecture and to understand the two types of input:


6. Defining model parameters:

The hyperparameters that we are going to use for our model is defined as follows:

If you know what this parameter means then you can play around it and can get better results.

Time Taken: I used the GPU on and hence it took me approximately 2 hours to train the model for 5 epochs. However, if you train it on a PC without GPU, it could take more time depending on the configuration of your machine.

7. Evaluating the model:

Since I have used the very small dataset for performing these experiments I am not able to get very good accuracy. The below code will calculate the accuracy of the model.


Since I have trained a model multiple times with different parameters you will not get the same accuracy as me. If you want you can directly download mode.h5 file from my google drive.


8. Final Thoughts:

One of the interesting thing about VQA is that it a completely new field. So there is absolutely no end to what you can do to solve this problem. Below are some tips while replicating the code.

  1. Start with a very small subset of data: When you start implementing I suggest you start with a very small amount of data. Because once you are ready with the whole setup then you can scale it any time.
  2. Understand the code: Understanding code line by line is very much helpful to match your theoretical knowledge. So for that, I suggest you can take very few samples(maybe 20 or less) and run a small chunk (2 to 3 lines) of code to get the functionality of each part.
  3. Be patient: One of the mistakes that I did while starting with this project was to do everything at one go. If you get some error while replicating code spend 4 to 5 days harder on that. Even after that if you won’t able to solve, I would suggest you resume after a break of 1 or 2 days. 

VQA is the intersection of NLP and CV and hopefully, this project will give you a better understanding (more precisely practically) with most of the deep learning concepts.

If you want to improve the performance of the model below are few tips you can try:

  1. Use larger datasets
  2. Try Building more complex models like Attention, etc
  3. Try using other pre-trained word embeddings like Glove 
  4. Try using a different architecture 
  5. Do more hyperparameter tuning

The list is endless and it goes on.

In the blog, I have not provided the complete code you can get it from my Github repository.

9. References:


DATANOMIQ MeetUp: Interactive Data Exploration and GUI’s in JupyterNotebooks

After our first successful collaboration Meetup with Mister Spex, we straightly continue with our next partner: VW Digital Labs!

Join us on Wednesday, October 9 for our DATANOMIQ Data Science Meetup at VW Digital Labs and get inspired.

Wednesday, October 9, time TBA

VW Digital Labs
Stralauer Allee 7, 10245 Berlin


18:30 doors open
19:00 Interactive Data Exploration and GUI’s in JupyterNotebooks – Christopher Kipp.
– using ipywidgets to get basic UI components and connet them
– qgrid to make Dataframes interactive (sortable, filterable, …)
– building interactive visualisations with bqplot

19:20 Q&A

10 minute break

19:40 second presentation
20:00 Q&A

20:15 networking


FREE ENTRY, snacks and drinks sponsored by VW digital labs.

Make sure to get your ticket:

Entrance only with registration.


Join our MeetUp group:

Zertifikatsstudium „Data Science and Big Data“ 2020 an der TU Dortmund

Jetzt bewerben!

Komplexe Daten aufbereiten und analysieren, um daraus zukünftige Entwicklungen abzulesen: das lernen Sie im berufsbegleitenden Zertifikatsstudium „Data Science and Big Data“ an der TU Dortmund.

Die Zielgruppe sind Fachkräfte, die sich in ihrer Berufspraxis mit Fragestellungen zum Thema Datenanalyse und Big Data befassen, jedoch nun tiefergehende Kenntnisse in dem Themenfeld erhalten möchten. Von der Analyse über das Management bis zur zielgerichteten Darstellung der Ergebnisse lernen die Teilnehmenden dabei Methoden der Disziplinen Statistik, Informatik und Journalistik kennen.

Renommierte Wissenschaftlerinnen und Wissenschaftler vermitteln den Teilnehmerinnen und Teilnehmern die neuesten datenwissenschaftlichen Erkenntnisse und zeigen, wie dieses Wissen praxisnah im eigenen Big-Data Projekt umgesetzt werden kann.

Die nächste Studiengruppe startet im Februar 2020, der Bewerbungsschluss ist am 4. November 2019. Die Anzahl der verfügbaren Plätze ist begrenzt, eine rechtzeitige Bewerbung lohnt sich daher.

Nähere Informationen finden Sie unter:

Dortmunder R-Kurse | Neue Termine im Herbst 2019

Erweitern Sie Ihre Fähigkeiten in der Anwendung der Open Source Statistiksoftware R: In der Tagesseminarreihe „Dortmunder R-Kurse“ an der Technischen Universität Dortmund geben erfahrene Wissenschaftler der Fakultät Statistik ihre Expertise an Sie weiter.

Sie erwerben dadurch Qualifikationen zur selbstständigen Analyse eigener Daten sowie Schlüsselkompetenzen im Umgang mit Big Data. Die Kurse richten sich an Anwenderinnen und Anwender jeder Fachrichtung aus Industrie und Forschungseinrichtungen, die ihre Daten mit R auswerten möchten.

Das Angebot umfasst Kurse für Einsteiger und Fortgeschrittene, wo Sie Ihre Kenntnisse in R erlernen und vertiefen können.

  • R Basiskurs
    Inhalte: Grundlagen zur ersten Datenanalyse
    Termine: 5. & 6. November 2019
  • R Vertiefungskurs
    Inhalt: Effiziente Analysen mit R
    Termine: 21. & 22. November 2019
  • Weitere Inhouse Themen auf Anfrage: Machine Learning in R, Shiny Apps mit R

Weitere Informationen zu den R-Kursen finden Sie unter:


Visual Question Answering with Keras – Part 1

This is Part I of II of the Article Series Visual Question Answering with Keras

Making Computers Intelligent to answer from images

If we look closer in the history of Artificial Intelligence (AI), the Deep Learning has gained more popularity in the recent years and has achieved the human-level performance in the tasks such as Speech Recognition, Image Classification, Object Detection, Machine Translation and so on. However, as humans, not only we but also a five-year child can normally perform these tasks without much inconvenience. But the development of such systems with these capabilities has always considered an ambitious goal for the researchers as well as for developers.

In this series of blog posts, I will cover an introduction to something called VQA (Visual Question Answering), its available datasets, the Neural Network approach for VQA and its implementation in Keras and the applications of this challenging problem in real life. 

Table of Contents:

1 Introduction

2 What is exactly Visual Question Answering?

3 Prerequisites

4 Datasets available for VQA

4.1 DAQUAR Dataset

4.2 CLEVR Dataset

4.3 FigureQA Dataset

4.4 VQA Dataset

5 Real-life applications of VQA

6 Conclusion


  1. Introduction:

Let’s say you are given a below picture along with one question. Can you answer it?

I expect confidently you all say it is the Kitchen without much inconvenience which is also the right answer. Even a five-year child who just started to learn things might answer this question correctly.

Alright, but can you write a computer program for such type of task that takes image and question about the image as an input and gives us answer as output?

Before the development of the Deep Neural Network, this problem was considered as one of the difficult, inconceivable and challenging problem for the AI researcher’s community. However, due to the recent advancement of Deep Learning the systems are capable of answering these questions with the promising result if we have a required dataset.

Now I hope you have got at least some intuition of a problem that we are going to discuss in this series of blog posts. Let’s try to formalize the problem in the below section.

  1. What is exactly Visual Question Answering?:

We can define, “Visual Question Answering(VQA) is a system that takes an image and natural language question about the image as an input and generates natural language answer as an output.”

VQA is a research area that requires an understanding of vision(Computer Vision)  as well as text(NLP). The main beauty of VQA is that the reasoning part is performed in the context of the image. So if we have an image with the corresponding question then the system must able to understand the image well in order to generate an appropriate answer. For example, if the question is the number of persons then the system must able to detect faces of the persons. To answer the color of the horse the system need to detect the objects in the image. Many of these common problems such as face detection, object detection, binary object classification(yes or no), etc. have been solved in the field of Computer Vision with good results.

To summarize a good VQA system must be able to address the typical problems of CV as well as NLP.

To get a better feel of VQA you can try online VQA demo by CloudCV. You just go to this link and try uploading the picture you want and ask the related question to the picture, the system will generate the answer to it.


  1. Prerequisites:

In the next post, I will walk you through the code for this problem using Keras. So I assume that you are familiar with:

  1. Fundamental concepts of Machine Learning
  2. Multi-Layered Perceptron
  3. Convolutional Neural Network
  4. Recurrent Neural Network (especially LSTM)
  5. Gradient Descent and Backpropagation
  6. Transfer Learning
  7. Hyperparameter Optimization
  8. Python and Keras syntax
  1. Datasets available for VQA:

As you know problems related to the CV or NLP the availability of the dataset is the key to solve the problem. The complex problems like VQA, the dataset must cover all possibilities of questions answers in real-world scenarios. In this section, I will cover some of the datasets available for VQA.

4.1 DAQUAR Dataset:

The DAQUAR dataset is the first dataset for VQA that contains only indoor scenes. It shows the accuracy of 50.2% on the human baseline. It contains images from the NYU_Depth dataset.

Example of DAQUAR dataset

Example of DAQUAR dataset

The main disadvantage of DAQUAR is the size of the dataset is very small to capture all possible indoor scenes.

4.2 CLEVR Dataset:

The CLEVR Dataset from Stanford contains the questions about the object of a different type, colors, shapes, sizes, and material.

It has

  • A training set of 70,000 images and 699,989 questions
  • A validation set of 15,000 images and 149,991 questions
  • A test set of 15,000 images and 14,988 questions

Image Source:


4.3 FigureQA Dataset:

FigureQA Dataset contains questions about the bar graphs, line plots, and pie charts. It has 1,327,368 questions for 100,000 images in the training set.

4.4 VQA Dataset:

As comapred to all datasets that we have seen so far VQA dataset is relatively larger. The VQA dataset contains open ended as well as multiple choice questions. VQA v2 dataset contains:

  • 82,783 training images from COCO (common objects in context) dataset
  • 40, 504 validation images and 81,434 validation images
  • 443,757 question-answer pairs for training images
  • 214,354 question-answer pairs for validation images.

As you might expect this dataset is very huge and contains 12.6 GB of training images only. I have used this dataset in the next post but a very small subset of it.

This dataset also contains abstract cartoon images. Each image has 3 questions and each question has 10 multiple choice answers.

  1. Real-life applications of VQA:

There are many applications of VQA. One of the famous applications is to help visually impaired people and blind peoples. In 2016, Microsoft has released the “Seeing AI” app for visually impaired people to describe the surrounding environment around them. You can watch this video for the prototype of the Seeing AI app.

Another application could be on social media or e-commerce sites. VQA can be also used for educational purposes.

  1. Conclusion:

I hope this explanation will give you a good idea of Visual Question Answering. In the next blog post, I will walk you through the code in Keras.

If you like my explanations, do provide some feedback, comments, etc. and stay tuned for the next post.

Erstellen und benutzen einer Geodatenbank

In diesem Artikel soll es im Gegensatz zum vorherigen Artikel Alles über Geodaten weniger darum gehen, was man denn alles mit Geodaten machen kann, dafür aber mehr darum wie man dies anstellt. Es wird gezeigt, wie man aus dem öffentlich verfügbaren Datensatz des OpenStreetMap-Projekts eine Geodatenbank erstellt und einige Beispiele dafür gegeben, wie man diese abfragen und benutzen kann.

Wahl der Datenbank

Prinzipiell gibt es zwei große “geo-kompatible” OpenSource-Datenbanken bzw. “Datenbank-AddOn’s”: Spatialite, welches auf SQLite aufbaut, und PostGIS, das PostgreSQL verwendet.

PostGIS bietet zum Teil eine einfachere Syntax, welche manchmal weniger Tipparbeit verursacht. So kann man zum Beispiel um die Entfernung zwischen zwei Orten zu ermitteln einfach schreiben:

während dies in Spatialite “nur” mit einer normalen Funktion möglich ist:

Trotztdem wird in diesem Artikel Spatialite (also SQLite) verwendet, da dessen Einrichtung deutlich einfacher ist (schließlich sollen interessierte sich alle Ergebnisse des Artikels problemlos nachbauen können, ohne hierfür einen eigenen Datenbankserver aufsetzen zu müssen).

Der Hauptunterschied zwischen PostgreSQL und SQLite (eigentlich der Unterschied zwischen SQLite und den meissten anderen Datenbanken) ist, dass für PostgreSQL im Hintergrund ein Server laufen muss, an welchen die entsprechenden Queries gesendet werden, während SQLite ein “normales” Programm (also kein Client-Server-System) ist welches die Queries selber auswertet.

Hierdurch fällt beim Aufsetzen der Datenbank eine ganze Menge an Konfigurationsarbeit weg: Welche Benutzer gibt es bzw. akzeptiert der Server? Welcher Benutzer bekommt welche Rechte? Über welche Verbindung wird auf den Server zugegriffen? Wie wird die Sicherheit dieser Verbindung sichergestellt? …

Während all dies bei SQLite (und damit auch Spatialite) wegfällt und die Einrichtung der Datenbank eigentlich nur “installieren und fertig” ist, muss auf der anderen Seite aber auch gesagt werden dass SQLite nicht gut für Szenarien geeignet ist, in welchen viele Benutzer gleichzeitig (insbesondere schreibenden) Zugriff auf die Datenbank benötigen.

Benötigte Software und ein Beispieldatensatz

Was wird für diesen Artikel an Software benötigt?

SQLite3 als Datenbank

libspatialite als “Geoplugin” für SQLite

spatialite-tools zum erstellen der Datenbank aus dem OpenStreetMaps (*.osm.pbf) Format

python3, die beiden GeoModule spatialite, folium und cartopy, sowie die Module pandas und matplotlib (letztere gehören im Bereich der Datenauswertung mit Python sowieso zum Standart). Für pandas gibt es noch die Erweiterung geopandas sowie eine praktisch unüberschaubare Anzahl weiterer geographischer Module aber bereits mit den genannten lassen sich eine Menge interessanter Dinge herausfinden.

– und natürlich einen Geodatensatz: Zum Beispiel sind aus dem OpenStreetMap-Projekt extrahierte Datensätze hier zu finden.

Es ist ratsam, sich hier erst einmal einen kleinen Datensatz herunterzuladen (wie zum Beispiel einen der Stadtstaaten Bremen, Hamburg oder Berlin). Zum einen dauert die Konvertierung des .osm.pbf-Formats in eine Spatialite-Datenbank bei größeren Datensätzen unter Umständen sehr lange, zum anderen ist die fertige Datenbank um ein vielfaches größer als die stark gepackte Originaldatei (für “nur” Deutschland ist die fertige Datenbank bereits ca. 30 GB groß und man lässt die Konvertierung (zumindest am eigenen Laptop) am besten über Nacht laufen – willkommen im Bereich “BigData”).

Erstellen eine Geodatenbank aus OpenStreetMap-Daten

Nach dem Herunterladen eines Datensatzes der Wahl im *.osm.pbf-Format kann hieraus recht einfach mit folgendem Befehl aus dem Paket spatialite-tools die Datenbank erstellt werden:

Erkunden der erstellten Geodatenbank

Nach Ausführen des obigen Befehls sollte nun eine Datei mit dem gewählten Namen (im Beispiel bremen-latest.sqlite) im aktuellen Ordner vorhanden sein – dies ist bereits die fertige Datenbank. Zunächst sollte man mit dieser Datenbank erst einmal dasselbe machen, wie mit jeder anderen Datenbank auch: Sich erst einmal eine Weile hinsetzen und schauen was alles an Daten in der Datenbank vorhanden und vor allem wo diese Daten in der erstellten Tabellenstruktur zu finden sind. Auch wenn dieses Umschauen prinzipiell auch vollständig über die Shell oder in Python möglich ist, sind hier Programme mit graphischer Benutzeroberfläche (z. B. spatialite-gui oder QGIS) sehr hilfreich und sparen nicht nur eine Menge Zeit sondern vor allem auch Tipparbeit. Wer dies tut, wird feststellen, dass sich in der generierten Datenbank einige dutzend Tabellen mit Namen wie pt_addresses, ln_highway und pg_boundary befinden.

Die Benennung der Tabellen folgt dem Prinzip, dass pt_*-Tabellen Punkte im Geokoordinatensystem wie z. B. Adressen, Shops, Bäckereien und ähnliches enthalten. ln_*-Tabellen enthalten hingegen geographische Entitäten, welche sich als Linien darstellen lassen, wie beispielsweise Straßen, Hochspannungsleitungen, Schienen, ect. Zuletzt gibt es die pg_*-Tabellen welche Polygone – also Flächen einer bestimmten Form enthalten. Dazu zählen Landesgrenzen, Bundesländer, Inseln, Postleitzahlengebiete, Landnutzung, aber auch Gebäude, da auch diese jeweils eine Grundfläche besitzen. In dem genannten Datensatz sind die Grundflächen von Gebäuden – zumindest in Europa – nahezu vollständig. Aber auch der Rest der Welt ist für ein “Wikipedia der Kartographie” insbesondere in halbwegs besiedelten Gebieten bemerkenswert gut erfasst, auch wenn nicht unbedingt davon ausgegangen werden kann, dass abgelegenere Gegenden (z. B. irgendwo auf dem Land in Südamerika) jedes Gebäude eingezeichnet ist.

Verwenden der Erstellten Datenbank

Auf diese Datenbank kann nun entweder direkt aus der Shell über den Befehl

zugegriffen werden oder man nutzt das gleichnamige Python-Paket:

Nach Eingabe der obigen Befehle in eine Python-Konsole, ein Jupyter-Notebook oder ein anderes Programm, welches die Anbindung an den Python-Interpreter ermöglicht, können die von der Datenbank ausgegebenen Ergebnisse nun direkt in ein Pandas Data Frame hineingeladen und verwendet/ausgewertet/analysiert werden.

Im Grunde wird hierfür “normales SQL” verwendet, wie in anderen Datenbanken auch. Der folgende Beispiel gibt einfach die fünf ersten von der Datenbank gefundenen Adressen aus der Tabelle pt_addresses aus:

Link zur Ausgabe

Es wird dem Leser sicherlich aufgefallen sein, dass die Spalte “Geometry” (zumindest für das menschliche Auge) nicht besonders ansprechend sowie auch nicht informativ aussieht: Der Grund hierfür ist, dass diese Spalte die entsprechende Position im geographischen Koordinatensystem aus Gründen wie dem deutlich kleineren Speicherplatzbedarf sowie der damit einhergehenden Optimierung der Geschwindigkeit der Datenbank selber, in binärer Form gespeichert und ohne weitere Verarbeitung auch als solche ausgegeben wird.

Glücklicherweise stellt spatialite eine ganze Reihe von Funktionen zur Verarbeitung dieser geographischen Informationen bereit, von denen im folgenden einige beispielsweise vorgestellt werden:

Für einzelne Punkte im Koordinatensystem gibt es beispielsweise die Funktionen X(geometry) und Y(geometry), welche aus diesem “binären Wirrwarr” den Längen- bzw. Breitengrad des jeweiligen Punktes als lesbare Zahlen ausgibt.

Ändert man also das obige Query nun entsprechend ab, erhält man als Ausgabe folgendes Ergebnis in welchem die Geometry-Spalte der ausgegebenen Adressen in den zwei neuen Spalten Longitude und Latitude in lesbarer Form zu finden ist:

Link zur Tabelle

Eine weitere häufig verwendete Funktion von Spatialite ist die Distance-Funktion, welche die Distanz zwischen zwei Orten berechnet.

Das folgende Beispiel sucht in der Datenbank die 10 nächstgelegenen Bäckereien zu einer frei wählbaren Position aus der Datenbank und listet diese nach zunehmender Entfernung auf (Achtung – die frei wählbare Position im Beispiel liegt in München, wer die selbe Position z. B. mit dem Bremen-Datensatz verwendet, wird vermutlich etwas weiter laufen müssen…):

Link zur Ausgabe

Ein Anwendungsfall für eine solche Liste können zum Beispiel Programme/Apps wie oder Google-Maps sein, in denen User nach Bäckereien, Geldautomaten, Supermärkten oder Apotheken “in der Nähe” suchen können sollen.

Diese Liste enthält nun alle Informationen die grundsätzlich gebraucht werden, ist soweit auch informativ und wird in den meißten Fällen der Datenauswertung auch genau so gebraucht, jedoch ist diese für das Auge nicht besonders ansprechend.

Viel besser wäre es doch, die gefundenen Positionen auf einer interaktiven Karte einzuzeichnen:

Was kann man sonst interessantes mit der erstellten Datenbank und etwas Python machen? Wer in Deutschland ein wenig herumgekommen ist, dem ist eventuell aufgefallen, dass sich die Endungen von Ortsnamen stark unterscheiden: Um München gibt es Stadteile und Dörfer namens Garching, Freising, Aubing, ect., rund um Stuttgart enden alle möglichen Namen auf “ingen” (Plieningen, Vaihningen, Echterdingen …) und in Berlin gibt es Orte wie Pankow, Virchow sowie eine bunte Auswahl weiterer *ow’s.

Das folgende Query spuckt gibt alle “village’s”, “town’s” und “city’s” aus der Tabelle pt_place, also Dörfer und Städte, aus:

Link zur Ausgabe

Graphisch mit matplotlib und cartopy in ein Koordinatensystem eingetragen sieht diese Verteilung folgendermassen aus:

Die Grafik zeigt, dass stark unterschiedliche Vorkommen der verschiedenen Ortsendungen in Deutschland (Clustering). Über das genaue Zustandekommen dieser Verteilung kann ich hier nur spekulieren, jedoch wird diese vermutlich ähnlichen Prozessen unterliegen wie beispielsweise die Entwicklung von Dialekten.

Wer sich die Karte etwas genauer anschaut wird merken, dass die eingezeichneten Landesgrenzen und Küstenlinien nicht besonders genau sind. Hieran wird ein interessanter Effekt von häufig verwendeten geographischen Entitäten, nämlich Linien und Polygonen deutlich. Im Beispiel werden durch die beiden Zeilen

die bereits im Modul cartopy hinterlegten Daten verwendet. Genaue Verläufe von Küstenlinien und Landesgrenzen benötigen mit wachsender Genauigkeit hingegen sehr viel Speicherplatz, da mehr und mehr zu speichernde Punkte benötigt werden (genaueres siehe hier).


Man kann also bereits mit einigen Grundmodulen und öffentlich verfügbaren Datensätzen eine ganze Menge im Bereich der Geodaten erkunden und entdecken. Gleichzeitig steht, insbesondere für spezielle Probleme, eine große Bandbreite weiterer Software zur Verfügung, für welche dieser Artikel zwar einen Grundsätzlichen Einstieg geben kann, die jedoch den Rahmen dieses Artikels sprengen würden.

A Bird’s Eye View: How Machine Learning Can Help You Charge Your E-Scooters

Bird scooters in Columbus, Ohio

Bird scooters in Columbus, Ohio

Ever since I started using bike-sharing to get around in Seattle, I have become fascinated with geolocation data and the transportation sharing economy. When I saw this project leveraging the mobility data RESTful API from the Los Angeles Department of Transportation, I was eager to dive in and get my hands dirty building a data product utilizing a company’s mobility data API.

Unfortunately, the major bike and scooter providers (Bird, JUMP, Lime) don’t have publicly accessible APIs. However, some folks have seemingly been able to reverse-engineer the Bird API used to populate the maps in their Android and iOS applications.

One interesting feature of this data is the nest_id, which indicates if the Bird scooter is in a “nest” — a centralized drop-off spot for charged Birds to be released back into circulation.

I set out to ask the following questions:

  1. Can real-time predictions be made to determine if a scooter is currently in a nest?
  2. For non-nest scooters, can new nest location recommendations be generated from geospatial clustering?

To answer these questions, I built a full-stack machine learning web application, NestGenerator, which provides an automated recommendation engine for new nest locations. This application can help power Bird’s internal nest location generation that runs within their Android and iOS applications. NestGenerator also provides real-time strategic insight for Bird chargers who are enticed to optimize their scooter collection and drop-off route based on proximity to scooters and nest locations in their area.


The electric scooter market has seen substantial growth with Bird’s recent billion dollar valuation  and their $300 million Series C round in the summer of 2018. Bird offers electric scooters that top out at 15 mph, cost $1 to unlock and 15 cents per minute of use. Bird scooters are in over 100 cities globally and they announced in late 2018 that they eclipsed 10 million scooter rides since their launch in 2017.

Bird scooters in Tel Aviv, Israel

Bird scooters in Tel Aviv, Israel

With all of these scooters populating cities, there’s much-needed demand for people to charge them. Since they are electric, someone needs to charge them! A charger can earn additional income for charging the scooters at their home and releasing them back into circulation at nest locations. The base price for charging each Bird is $5.00. It goes up from there when the Birds are harder to capture.

Data Collection and Machine Learning Pipeline

The full data pipeline for building “NestGenerator”


From the details here, I was able to write a Python script that returned a list of Bird scooters within a specified area, their geolocation, unique ID, battery level and a nest ID.

I collected scooter data from four cities (Atlanta, Austin, Santa Monica, and Washington D.C.) across varying times of day over the course of four weeks. Collecting data from different cities was critical to the goal of training a machine learning model that would generalize well across cities.

Once equipped with the scooter’s latitude and longitude coordinates, I was able to leverage additional APIs and municipal data sources to get granular geolocation data to create an original scooter attribute and city feature dataset.

Data Sources:

  • Walk Score API: returns a walk score, transit score and bike score for any location.
  • Google Elevation API: returns elevation data for all locations on the surface of the earth.
  • Google Places API: returns information about places. Places are defined within this API as establishments, geographic locations, or prominent points of interest.
  • Google Reverse Geocoding API: reverse geocoding is the process of converting geographic coordinates into a human-readable address.
  • Weather Company Data: returns the current weather conditions for a geolocation.
  • LocationIQ: Nearby Points of Interest (PoI) API returns specified PoIs or places around a given coordinate.
  • OSMnx: Python package that lets you download spatial geometries and model, project, visualize, and analyze street networks from OpenStreetMap’s APIs.

Feature Engineering

After extensive API wrangling, which included a four-week prolonged data collection phase, I was finally able to put together a diverse feature set to train machine learning models. I engineered 38 features to classify if a scooter is currently in a nest.

Full Feature Set

Full Feature Set

The features boiled down into four categories:

  • Amenity-based: parks within a given radius, gas stations within a given radius, walk score, bike score
  • City Network Structure: intersection count, average circuity, street length average, average streets per node, elevation level
  • Distance-based: proximity to closest highway, primary road, secondary road, residential road
  • Scooter-specific attributes: battery level, proximity to closest scooter, high battery level (> 90%) scooters within a given radius, total scooters within a given radius


Log-Scale Transformation

For each feature, I plotted the distribution to explore the data for feature engineering opportunities. For features with a right-skewed distribution, where the mean is typically greater than the median, I applied these log transformations to normalize the distribution and reduce the variability of outlier observations. This approach was used to generate a log feature for proximity to closest scooter, closest highway, primary road, secondary road, and residential road.

An example of a log transformation

Statistical Analysis: A Systematic Approach

Next, I wanted to ensure that the features I included in my model displayed significant differences when broken up by nest classification. My thinking was that any features that did not significantly differ when stratified by nest classification would not have a meaningful predictive impact on whether a scooter was in a nest or not.

Distributions of a feature stratified by their nest classification can be tested for statistically significant differences. I used an unpaired samples t-test with a 0.01% significance level to compute a p-value and confidence interval to determine if there was a statistically significant difference in means for a feature stratified by nest classification. I rejected the null hypothesis if a p-value was smaller than the 0.01% threshold and if the 99.9% confidence interval did not straddle zero. By rejecting the null-hypothesis in favor of the alternative hypothesis, it’s deemed there is a significant difference in means of a feature by nest classification.

Battery Level Distribution Stratified by Nest Classification to run a t-test

Battery Level Distribution Stratified by Nest Classification to run a t-test

Log of Closest Scooter Distribution Stratified by Nest Classification to run a t-test

Throwing Away Features

Using the approach above, I removed ten features that did not display statistically significant results.

Statistically Insignificant Features Removed Before Model Development

Model Development

I trained two models, a random forest classifier and an extreme gradient boosting classifier since tree-based models can handle skewed data, capture important feature interactions, and provide a feature importance calculation. I trained the models on 70% of the data collected for all four cities and reserved the remaining 30% for testing.

After hyper-parameter tuning the models for performance on cross-validation data it was time to run the models on the 30% of test data set aside from the initial data collection.

I also collected additional test data from other cities (Columbus, Fort Lauderdale, San Diego) not involved in training the models. I took this step to ensure the selection of a machine learning model that would generalize well across cities. The performance of each model on the additional test data determined which model would be integrated into the application development.

Performance on Additional Cities Test Data

The Random Forest Classifier displayed superior performance across the board

The Random Forest Classifier displayed superior performance across the board

I opted to move forward with the random forest model because of its superior performance on AUC score and accuracy metrics on the additional cities test data. AUC is the Area under the ROC Curve, and it provides an aggregate measure of model performance across all possible classification thresholds.

AUC Score on Test Data for each Model

AUC Score on Test Data for each Model

Feature Importance

Battery level dominated as the most important feature. Additional important model features were proximity to high level battery scooters, proximity to closest scooter, and average distance to high level battery scooters.

Feature Importance for the Random Forest Classifier

Feature Importance for the Random Forest Classifier

The Trade-off Space

Once I had a working machine learning model for nest classification, I started to build out the application using the Flask web framework written in Python. After spending a few days of writing code for the application and incorporating the trained random forest model, I had enough to test out the basic functionality. I could finally run the application locally to call the Bird API and classify scooter’s into nests in real-time! There was one huge problem, though. It took more than seven minutes to generate the predictions and populate in the application. That just wasn’t going to cut it.

The question remained: will this model deliver in a production grade environment with the goal of making real-time classifications? This is a key trade-off in production grade machine learning applications where on one end of the spectrum we’re optimizing for model performance and on the other end we’re optimizing for low latency application performance.

As I continued to test out the application’s performance, I still faced the challenge of relying on so many APIs for real-time feature generation. Due to rate-limiting constraints and daily request limits across so many external APIs, the current machine learning classifier was not feasible to incorporate into the final application.

Run-Time Compliant Application Model

After going back to the drawing board, I trained a random forest model that relied primarily on scooter-specific features which were generated directly from the Bird API.

Through a process called vectorization, I was able to transform the geolocation distance calculations utilizing NumPy arrays which enabled batch operations on the data without writing any “for” loops. The distance calculations were applied simultaneously on the entire array of geolocations instead of looping through each individual element. The vectorization implementation optimized real-time feature engineering for distance related calculations which improved the application response time by a factor of ten.

Feature Importance for the Run-time Compliant Random Forest Classifier

Feature Importance for the Run-time Compliant Random Forest Classifier

This random forest model generalized well on test-data with an AUC score of 0.95 and an accuracy rate of 91%. The model retained its prediction accuracy compared to the former feature-rich model, but it gained 60x in application performance. This was a necessary trade-off for building a functional application with real-time prediction capabilities.

Geospatial Clustering

Now that I finally had a working machine learning model for classifying nests in a production grade environment, I could generate new nest locations for the non-nest scooters. The goal was to generate geospatial clusters based on the number of non-nest scooters in a given location.

The k-means algorithm is likely the most common clustering algorithm. However, k-means is not an optimal solution for widespread geolocation data because it minimizes variance, not geodetic distance. This can create suboptimal clustering from distortion in distance calculations at latitudes far from the equator. With this in mind, I initially set out to use the DBSCAN algorithm which clusters spatial data based on two parameters: a minimum cluster size and a physical distance from each point. There were a few issues that prevented me from moving forward with the DBSCAN algorithm.

  1. The DBSCAN algorithm does not allow for specifying the number of clusters, which was problematic as the goal was to generate a number of clusters as a function of non-nest scooters.
  2. I was unable to hone in on an optimal physical distance parameter that would dynamically change based on the Bird API data. This led to suboptimal nest locations due to a distortion in how the physical distance point was used in clustering. For example, Santa Monica, where there are ~15,000 scooters, has a higher concentration of scooters in a given area whereas Brookline, MA has a sparser set of scooter locations.

An example of how sparse scooter locations vs. highly concentrated scooter locations for a given Bird API call can create cluster distortion based on a static physical distance parameter in the DBSCAN algorithm. Left:Bird scooters in Brookline, MA. Right:Bird scooters in Santa Monica, CA.

An example of how sparse scooter locations vs. highly concentrated scooter locations for a given Bird API call can create cluster distortion based on a static physical distance parameter in the DBSCAN algorithm. Left:Bird scooters in Brookline, MA. Right:Bird scooters in Santa Monica, CA.

Given the granularity of geolocation scooter data I was working with, geospatial distortion was not an issue and the k-means algorithm would work well for generating clusters. Additionally, the k-means algorithm parameters allowed for dynamically customizing the number of clusters based on the number of non-nest scooters in a given location.

Once clusters were formed with the k-means algorithm, I derived a centroid from all of the observations within a given cluster. In this case, the centroids are the mean latitude and mean longitude for the scooters within a given cluster. The centroids coordinates are then projected as the new nest recommendations.

NestGenerator showcasing non-nest scooters and new nest recommendations utilizing the K-Means algorithm

NestGenerator showcasing non-nest scooters and new nest recommendations utilizing the K-Means algorithm.

NestGenerator Application

After wrapping up the machine learning components, I shifted to building out the remaining functionality of the application. The final iteration of the application is deployed to Heroku’s cloud platform.

In the NestGenerator app, a user specifies a location of their choosing. This will then call the Bird API for scooters within that given location and generate all of the model features for predicting nest classification using the trained random forest model. This forms the foundation for map filtering based on nest classification. In the app, a user has the ability to filter the map based on nest classification.

Drop-Down Map View filtering based on Nest Classification

Drop-Down Map View filtering based on Nest Classification

Nearest Generated Nest

To see the generated nest recommendations, a user selects the “Current Non-Nest Scooters & Predicted Nest Locations” filter which will then populate the application with these nest locations. Based on the user’s specified search location, a table is provided with the proximity of the five closest nests and an address of the Nest location to help inform a Bird charger in their decision-making.

NestGenerator web-layout with nest addresses and proximity to nearest generated nests

NestGenerator web-layout with nest addresses and proximity to nearest generated nests


By accurately predicting nest classification and clustering non-nest scooters, NestGenerator provides an automated recommendation engine for new nest locations. For Bird, this application can help power their nest location generation that runs within their Android and iOS applications. NestGenerator also provides real-time strategic insight for Bird chargers who are enticed to optimize their scooter collection and drop-off route based on scooters and nest locations in their area.


The code for this project can be found on my GitHub

Comments or Questions? Please email me an E-Mail!