MishMash hackathon is India’s largest online diversity hackathon. The focus will be to give you, regardless of your background, gender, sexual orientation, ethnicity, age, skill sets and viewpoints, an opportunity to showcase your talent. The Hackathon is Live from 6:00 PM, 23rd March to 11:55 PM, 1st April, 2020
2
stars
12
forks
source link
DataRenaissance - Unilever Data Science POC Use Case - TH Deep Tech/Machine Learning Theme EME #134
Deep Tech - Problem Statement - 3: Yes, Submitted both dataset solution
Deep Tech - Problem Statement - 2: NA
Azure Services Used- Kindly mention the Azure Services used in your project.
Azure services I would use are deployment, testing, monitoring and chatbot services
🔥 Your Pitch
First I have extracted some faetures such as Year and period. For imputation I have used mean values for the feature.
I have used regression models such Linear Regression, Ridge, Lasso, RandomForest and XGBoost Regressor.
In case of Lasso there were faetures which were of no importance and their coefficients were reduced to zero. Thus implies dimensionality reduction and much simple model.
Lastly I have tested the model on Ridge which gave best MAPE for the given dataset and for test data I got MAPE of 14.34
DataRenaissance - Unilever Data Science POC Use Case - TH Deep Tech/Machine Learning Theme EME
ℹ️ Project information
Project Name: Sales Predictions
Short Project Description: Predicting Sales using Machine Learning
Team Name: DataRenaissance
Team Members: Gaurav Kantrod https://github.com/gauravkantrod
Demo Link: NA
Repository Link(s): https://github.com/gauravkantrod/Skillenza-MishMash-MachineLearning
Presentation Link: NA
Deep Tech - Problem Statement - 3: Yes, Submitted both dataset solution
Deep Tech - Problem Statement - 2: NA
Azure Services Used- Kindly mention the Azure Services used in your project. Azure services I would use are deployment, testing, monitoring and chatbot services
🔥 Your Pitch
First I have extracted some faetures such as Year and period. For imputation I have used mean values for the feature. I have used regression models such Linear Regression, Ridge, Lasso, RandomForest and XGBoost Regressor. In case of Lasso there were faetures which were of no importance and their coefficients were reduced to zero. Thus implies dimensionality reduction and much simple model. Lastly I have tested the model on Ridge which gave best MAPE for the given dataset and for test data I got MAPE of 14.34
Data_Renaissance_MishMash_Gaurav_Kantrod.zip
🔦 Any other specific thing you want to highlight?
(Optional) I have made a dataframe for LASSO which have some coefficients equal to 0 which show the feature selection technique.
✅ Checklist
Before you post the issue: