This project is based on the Digit Recognizer competition on Kaggle, found here: https://www.kaggle.com/c/digit-recognizer/data.
To sum it up in a nutshell, it is about predicting the correct handwritten digit based on its image pixels. The dataset is the famous MNIST-dataset by Yann LeCun at al. (http://yann.lecun.com/exdb/mnist/).
There are several ways to solve this "problem" (for example with a K-Nearest-Neighbor algorithm, found in my repository here: https://github.com/IIIskiplikIII/another-MNIST-try).
This time instead I chose the Deep Neural Network approach with a Convolutional Neural Network.
Based on the MNIST Digit Recognition from Kaggle I tried several different approaches / models / algorithms to gain a closer look into their different behaviours and become more familiar with their usage. You can find them as well on Kaggle and here on GitHub. They mostly look the same, especially in the preprocessing part from plain csv file to dataframe or other, for a model usable data formats. The focus here is on the model part themself so do not wonder that some parts look equal.