This repo contains series of notebooks, which are my submissions to a online Hack-a-thon : Code-2-ML hosted by CodeChef Chapter VIT AP and ML Club VIT AP.
Initially I was given two datasets namely Train_dataset_ - Train_Aug10.csv and Test_dataset - Put-Call_TS.csv.
This dataset contains various stock's fractors (features) and the respective stock price. The Problem is to make an Machine Learning Regression Model that first learns on the training dataset and then predicts the value for the test dataset. The predicted .csv file is saved as file_01.csv
Algorithm used : Random Forest Regressor
This dataset contains a single stock's PutCall Ratio parameter, but on 5 different consecutive days. The Problem is to make an efficient Machine Learning Time Series Model that first learns on the training dataset and then predicts the value on test dataset. The predicted .csv file is saved as file_02.csv
Algorithm used : VAR : Vector Auto Regressor
You can also avoid all the below steps and instead load the notebooks on Google Colab appropriately with the datasets.
pip install -r requirements.txt #Preferable to install on a Virtual Environment If OSx / Linux use pip3 instead of pip
There are 2 Notebooks names problem_01.ipynb and test_dataprep.ipynb.
problem_01.ipynb is the actual problem notebook.
test_dataprep.ipynb is to clean the test Dataset. Which is then exported to be used by the main Notebook.
This all could have avoided and made into a single Notebook but it was clumsy. So for better readability, we had to make two different Notebook.
Run the problem_02.ipynb and the solution will be outputted. And respective csv file is save in Answer_files directory as file_02.csv.
After the execution of both the notebooks the .csv file solutions wil be exported.
These .csv files (solutions) are present into the Answer_files.
This Project has secured 1st Place 🥇 in the Code-2-ML Hack-a-thon.
This repo contains other participant's submissions for the Hack-a-thon.
