No description, website, or topics provided.
Python
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
LICENSE
README.md
convert_input_files.ipynb
debt.png
frank_lampard.jpg
google_prediction_fft_ave.ipynb
google_prediction_fft_ave_everything.ipynb
google_prediction_fft_everything.ipynb
googleprediction.py
ipython_fft_example.ipynb
ml_with_goog_pred.ipynb
sklearn_fft_ave.ipynb
sklearn_fft_ave_everything.ipynb
snow.jpg
train.csv
train_cleaned.csv

README.md

Utah Code Camp 2015

IPython notebooks and supporting files for a presentation on the Google Prediction API. The primary result in the presentation is a performance benchmark of the Google Prediction API against sklearn.RandomForestClassifier. Key files include:

  • ml_with_goog_pred.ipynb: The main presentation. A brief introduction to machine learning, an example of predicting Titanic survival using the Google Prediciton API, and then a comparitive benchmark between sklearn.RandomForestClassifier and the Google Prediction API. The benchmark was conducted as part of a data science competition to classify very short Taylor Swift audio clips as either huge hit or not hit.
  • sklearn_fft_ave.ipynb: A notebook that details the RamdomForestClassifier benchmarking solution, including the feature engineering to compute frequency spectra.
  • google_prediction_fft_ave.ipynb: A notebook that details the Google Prediction API benchmark solution.
  • googleprediction.py: A simple wrapper around the Google Predication API functions to make it look more like the scikit-learn standard interface to models.
  • ipython_fft_example.ipynb: A notebook showing a quick example of computing FFTs in scientific Python. The emphasis is on preserving the physical significance of the resulting frequency spectrum.