Titanic: Machine Learning from Disaster

Posed a question about a dataset, then used NumPy and Pandas to answer that question based on the data and created a report to share the results.

What have I learned?

This was my first major project in the field of data science. Here I gained my first experiences from data exploration to the presentation of results. Here I also learned to love this subject!

Bike Sharing Dataset

In this project, I build a neural network from scratch (only in numpy). I used this neural network to predict a load of a bicycle rental system in the city

What have I learned?

Since I only used Numpy, I got a very good understanding of gradient descent, backpropagation and other concepts in deep leaning. These are all important for using Tensorflow as an effective tool.

Image Classification from the CIFAR-10 dataset

This dataset consists of different classes of pictures, e.g. airplanes, dogs and cats. The first task in this project was to preprocess the input data. As a second step, a CNN should be built that should correctly classify the respective object on the image.

What have I learned?

I learned how to use the library Tensorflow. I also learned how to classify images with a neural network. This includes the step of normalizing the input data, one-hot encode the labels, build a convolutional layer, max pool layer, and fully connected layer.

Creating and designing my portfolio website

Data Science Intern

Building a data analysis Tool from scratch to analyzing driving data and deriving an understanding of links between technologies, environment and customer needs, to simulate buying decisions.

What have I learned?

By Building the Tools i used the Python libraries: NumPy, pandas, SciPy, matplotlib, and scikit-learn

Used GitLab for version control

Used Scrum Methode for agile project management

Working Student Data Science

Optimization of a data analysis Tool to analyzing driving data and deriving an understanding of links between technologies, environments and customer needs, to simulate buying decisions. Make Costumer Analysis Tool 38 % faster with parallel computing library dask. Optimise user-friendliness by issuing warnings in the event of incorrect entries.

What have I learned?

How to use dask for parallel computing in python

Comparison of common machine learning concepts in the network integration of renewable energies

Creating a presentation about the common machine learning algorithm in renewable energy.

What have I learned?

I learned a lot about the different concepts in machine learning and learn how present this complex stuff easy, e.g. Bayesian Algorithm's, Decision Trees, Support Vector Machine, k-nearest-neighbour, artificial neural networks

TV Script Generation

Text generation with a recurrent neural network.

What have I learned?

The concept of recurrent neural networks