A personal repository containing some small data science related projects with the goal of learning some interesting techniques and algorithms.
The tool of choice is Python using Scikit-Learn and Tensorflow APIs in Jupyter notebooks.
If GitHub is having trouble loading the notebook, you can view the notebook by going to Jupyter's nbviewer.
Highlights from Google's Machine Learning Crash Course
- Predict movies as favorable using text reviews done with an embedding and a deep neural network
- MNIST digit classification using a multilayer perceptron neural network: Google Course Implementation
- Image classification to distinguish between images of cats and dogs using a convolutional neural network and fine-tuning of the Inception_v3 network
Highlights from the O'Reilly Machine Learning Book:
- Predicting real estate values including data pipelines and hyperparameter searching.
- An in-depth look at training models including linear, logistic, and softmax regression with gradient descent and direct parameter optimization
- Classification training with MNIST data
- MNIST digit classification using a multilayer perceptron neural network: O'Reilly Implementation
- An overview of common Dimensionality Reduction techniques
- Ensemble modeling including random forests and boosting
- Fun with CNNs including feature extraction, MNIST classification, and using Google's Inception_v3 network to make predictions
- Image classification using a convolutional neutral network to distinguish between 6 sets of pictures: glaciers, mountains, streets, buildings, forest, or sea
- German to English translation using an encoder-decoder network with word embedding and Gated Recurrent Unit (GRU) cells.
- KMeans clustering on a small set of mall customer data
- Diabetes prediction using three different types of supervised learning algorithms: k-Nearest neighbor, random forest, and naive bayes
- Investigation of Suicide Data from around the world with pandas manipulation and seaborn plotting