Machine Learning Challenge - Chaordic

Machine Learning Challenge from Chaordic

Steps to solve the challenge

Gathering information: research similar problems and published articles
Knowing the data: play with the datasets and see how it is structured
Cleaning the data: split the dataset and remove irrelevant information
Analysing the data: find correlations between possible features and the target data
Training the model: split the data for train and test and try different models of classifiers
Evaluating the model: compare the results with an appropriate metric for a binary classification
Submitting the answer: generate csv file with the target data scored with the best trained model
Iterating over the process: go back to step 4, filter the relevant features, remove outliers, balance the trainig data, combine some variables, do some magic, ...

Gathering information: google
Knowing the data: gedit and bash commands
Cleaning the data: bash scripts and OpenRefine
Analysing the data: jupyter notebooks and python libraries (pandas, numpy, matplotlib, ...)
Training the model: sklearn modules
Evaluating the model: sklearn functions and methods inside classes
Submitting the answer: python code
Iterating over the process: try different tools like azure machine learning studio

data/ : directory to store the datasets
- data/split_data.sh : bash script to split dataset
  - e.g.:$./split_data.sh data
- data/stats.sh : bash script to count masculine/feminine gender
  - e.g.:$./stats.sh data
- data/test* : file with samples of the datasets' content
- data/users : all target users ids
research/ : directory with articles related to the challenge
playground.ipynb : notebook used to learn how to use python packages (pandas, sklearn, ...)
Igor_final.ipynb : notebook with the workflow of the solution.
final_answer.csv : file containing challenge's answer