Build ML Classifier and Yield Dashboard using Python
Budget: $750 – $1,500 USD
The aim of this project is to find fast and accurate yield prediction model based on high volume manufacturing (HVM) process yield dataset.
The yield data contains all process parametric and electrical test yield data. Based on process data , we are trying to predict the fallout electrical test bin while also identifying the possible process which may contribute to the loss. From this project we also target to build a descriptive visual analytic model for easy data exploration ( Ex: Trend chart, Column bar, Box plot)
Objective
• To determine best yield prediction machine learning model which can provide best result in terms of defined success metric using Intel manufacturing process.
• To provide best descriptive analytics visualization method using best yield model that would allow for user easy data exploration, analysis allowing for quick decision-making process.
• To evaluate the performance of yield prediction models to identify the best model.
Project Requirement
- Data pre-processing to remove empty data column and correct format columns
-Perform feature selection using Lasso Regression and Information Gain to eliminate unwanted or less important features
-Use SVM, GBDT and Random Forest ML algorithm to create yield prediction model ( Class Label highlighted yellow in raw data)
-Perform evaluation using Accuracy, AUC-ROC and TPR(true-positive rate)
-Build descriptive visual dashboard which will highlight important information from yield data using visual idioms
Raw Data file is exceeding 100MB and will be provided upon project confirmation
The yield data contains all process parametric and electrical test yield data. Based on process data , we are trying to predict the fallout electrical test bin while also identifying the possible process which may contribute to the loss. From this project we also target to build a descriptive visual analytic model for easy data exploration ( Ex: Trend chart, Column bar, Box plot)
Objective
• To determine best yield prediction machine learning model which can provide best result in terms of defined success metric using Intel manufacturing process.
• To provide best descriptive analytics visualization method using best yield model that would allow for user easy data exploration, analysis allowing for quick decision-making process.
• To evaluate the performance of yield prediction models to identify the best model.
Project Requirement
- Data pre-processing to remove empty data column and correct format columns
-Perform feature selection using Lasso Regression and Information Gain to eliminate unwanted or less important features
-Use SVM, GBDT and Random Forest ML algorithm to create yield prediction model ( Class Label highlighted yellow in raw data)
-Perform evaluation using Accuracy, AUC-ROC and TPR(true-positive rate)
-Build descriptive visual dashboard which will highlight important information from yield data using visual idioms
Raw Data file is exceeding 100MB and will be provided upon project confirmation