I can help you to..

---

• Harvest data from different sources like SQL databases or messy Excel spreadsheets.
• Reformat and reshape data to make further analyses easier.
• Clean datasets removing outliers and abberant data.

Example:

---

• I created a dashboard for the municipality of Oosterhout. It is directly plugged on their SQL database and allows to access its information in real time.

Contact me

---

Tell me more about your project at anouk.okkema@gmail.com to see if we can work together.

I can help you for

---

• Exploratory analysis of a dataset.
• Creation of clean and reproducible reports using Google Colab.
• Application of statistical methods on your dataset.

Example:

---

• I'm the author of several projects where I was in charge to explore and understand complex dataset in order to answer scientific questions. The projects are viewed in the portfolio section.

Contact me

---

Tell me more about your project at anouk.okkema@gmail.com to see if we can work together.

I can help you with:

---

• Creation and review of graphs for your project.
• Dashboard creation for real time data analysis using Power BI
• Visualization of geospatial data.

Example:

---

• See examples in the portfolio section of this website.

Contact me

---

Tell me more about your project at anouk.okkema@gmail.com to see if we can work together.

Description

---

The goal of the dashboard is to explore one specific role within the theme of digitalisation and help the municipality of Oosterhout improve their business processes using data. One of the key responsibilities of the municipality of Oosterhout (henceforth the client) is youth care. While Dutch children are among the happiest and healthiest in the world, there are still many children that need extra support and care.

To ensure a youth care system that is more efficient, coherent, and cost-effective, the client has chosen to digitalize their existing youth care process. The client has approached me – the data analyst – with the aim to discover more efficient (in cost and time) ways of delivering youth care. The dashboard contains a research question, and the dashboard pages contain the data which I received to answer the research question.

This dashboard is going to research the money that is left over.

The dashboard application allows the user to see the differences between women and men when receiving money and invoices. The money left over is calculated via a DAX measure that contains two variables, the allotment amount and the invoice amount. The invoice amount was substracted from the allotment amount to calculate the money left over.

Read more

---

The dashboard application is available online. If you want to know more about the dashboard, you can read the dashboard manual which explains the features of the dashboard. The source code of the project are of course on Github. The source code contains the EDA and implementations of Machine Learning on the data from the Power BI dashboard.

Description

---

The prototype will be a demonstration of an application in which the computer vision algoritm is used as a fundamental feature. The problem my algorithm solves is given a set of car images and prior knowledge the user should be able to assign the label car to the images.

But what if you have a folder containing the images of cars, statues, dogs, mammals, and keyboards, and they are just placed random in the folder. Now imagine someone gives you a possible valid image of a mammal, and you want to look it up in the dataset if it is correct. How would you do it?

The algorithm will be able to differentiate between a mammal, a dog, a car, a keyboard, and a statue. The algorithm will assign one of the labels (car, dog, keyboard, mammal, and statue) to the image through training previous images with a label. The algorithm analyses the visible characteristics of the images and learns about the classes.

The Algorithm of Image Classification

---

When an image without a label comes as input into the model, the model analyses the visible characteristics (sex, dog fur colour, car colour, eye colour, etc.) of the image, and put a predicted label on it. The algorithm shows the confidence it has in being correct.

If the user wants to know how the algorithm interprets the image, the algorithm scanned the image through Grad-CAMs and shows the image with visual explanations about how the algorithm knows which label needs to be assigned to the image.

My algorithm contains a CNN network and has shown excellent results regarding image classification. Therefore, my algorithm can identify the images without labels.

Screens

---

Want to know more about the screens in the video? Well you're in luck! If you hover over the images, you will get explained what the purpose of that specific screen is.

Read more

---

The prototype is based on my algorithm. The algorithm contains deep neural networks that can accurately classify an image while being sensitive to biases that may be present in the data the network is trained on. The algorithm had undergo a few steps:

1. Data Exploration: Explore the Google Open Images dataset and discover biases if any are present
2. Baseline Algorithm Design: Design and train a suitable baseline deep learning architecture
3. Baseline Algorithm Improvement: Tune the hyperparameters of the model and use data pre-processing and data augmentation techniques if necessary to improve performance
4. SOTA comparison: Compare the baseline models with SOTA (Transfer Learning & Fine Tuning)
5. XAI: Explain the results of the AI algorithm and evaluate it's fairness, transparency, and interpretability

The source code of the algorithm is on GitHub.

If you are wondering about the web application, it is not available online. The application was made in a free trial of proto.io which can't be activated again.

Description

---

The municipality of Breda examined the state of segregation in Breda and the extent to which Breda's policy is effective in combating segregation in the city. There are several indications that segregation and social inequality between neighborhoods have increased in the past 10 years. I will use the skills and knowledge I have gained over the course of the first year to build an example of how to cluster the neighborhoods for the City of Breda to help combat rising segregation in a completely data-driven manner.

I created a dashboard allowing to explore data from Breda to cluster the neighborhoods with the DEC method. It allows to quickly understand which neighborhoods are similar to each other and clustered within the same cluster.

The dashboard was created using Power BI. Here is a screenshot of the application:

Deep Embedded Clustering

---

Recently a Deep Embedded Clustering (DEC) method was published. It combines autoencoder with K-means and other machine learning techniques for clustering rather than dimensionality reduction.

Deep Embedded Clustering (DEC) is a method that simultaneously learns feature representations and cluster assignments using deep neural networks. DEC learns a mapping from the data space to a lower-dimensional feature space in which it iteratively optimizes a clustering objective.

One of the key components in the DEC method is soft labeling, that is, assigning an estimated class to each of the data samples in such a way that it can be refined iteratively. To this end, a new ClusteringLayer class is defined which converts an input sample (aka feature) to a soft label.

The Deep Embedded Clustering method was used for this project. The source code of the project are of course on Github. The source code contains the Deep Embedded Clustering method to use the data from Breda to cluster the neighborhoods of Breda. It was implemented in the Power BI dashboard.

Read more

---

I can't provide more information on this project for privacy reasons.

The Deep Clustering Algorithm is on Github.

Description

---

The poster is answering the main research question: What is the correlation between economic income from Brazil and having the possibility to have access to basic services?

To answer the question, we first must state the problem: The world as we know it now, does not have basic services in all the countries. We are moving forward with equality for basic services. But we are still not there. In order to provide basic services at a minimal cost, the world needs water sources, electricity sources, sanitation facilities to be improved and cooking fuels and technologies to be clean all around the world. However, the procedures should not damage the nature.

Now why is it a problem? Every country has not (enough) basic services what will cause famine, poor people, bad water and more. If we have basic services all around the world, the world will change in a good way.

At last, how will my question provide insight into the problem? The question will show the correlation between economic income from a country and having the possibility to have access to basic services. It will show solutions to the causes and the effects of the basic services in Brazil.

Read more

---

The poster is visualized using static graphs with R.