Building a Machine Learning Model to Predict the Sales of Paints Manufacturer

About Sales prediction using Custom ML model
Industry:
Business Type:
*********
Our Process
1
Conceptualization
Competition Analysis
Customer Data Analysis
Customer Data Analysis
2
Design
User Journey Mapping
App Design Improvement
App Design Improvement
3
Development
Android App Development
iOS App Development
iOS App Development
4
Deployment
App Store & Play Store Deployment
Post-Deployment Support
Post-Deployment Support

Our Project Challenges
1
Predicting New and Prototype Part Sales
Forecasting sales for newly manufactured parts and prototypes posed a significant challenge due to the lack of historical data. To address this, we leveraged relevant characteristics shared between existing parts and the prototypes.
2
Skewed Data from APIs
The client’s data, obtained through APIs, exhibited skewness. This skewed data significantly impacted model training. To mitigate this, our team developed an API to extract more current and balanced data, ensuring better model performance.
Tools and Technologies Used

The Results
For SKUs with fewer than 1000 units in sales, the prediction variance ranged between -100 and +100. For SKUs selling over 1000 units, the variance was within -20% to +20%. After deploying the machine learning model, we continue to provide supervised training using updated data