File size: 2,422 Bytes
e5844bd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 |
---
license: apache-2.0
---
Model Card: TERPredictor V1
📌 Model Name
TERPredictor V1 – A linear regression model for predicting the Total Expense Ratio (TER) of mutual fund Regular Plans.
📖 Overview
TERPredictor V1 is a regression model trained to estimate the 'Regular Plan - Total TER (%)' of mutual funds based on various financial features. It uses a simple linear regression approach and achieves near-perfect performance on the test set. Due to the unusually high accuracy, this model is best suited for exploratory analysis and feature relationship interpretation, rather than generalization to unseen data.
📊 Intended Uses
Expense Ratio Estimation: Estimate TER for new or hypothetical mutual fund structures.
Outlier Detection: Identify funds with unusually high or low TERs.
Feature Impact Analysis: Understand which components most influence TER.
🧠 Model Architecture
Attribute Value
Model Type Linear Regression
Framework scikit-learn
Input Features 10 float64 columns
Target Variable Regular Plan - Total TER (%)
Identifier Dropped Scheme Name (object)
📚 Training Details
Dataset Size: 1,622 samples
Train/Test Split: 1297 / 325
Missing Values: None
Preprocessing:
Dropped identifier column (Scheme Name)
No normalization required due to linear model simplicity
📈 Evaluation Metrics
Metric Value
Mean Squared Error (MSE) 0.000001
R-squared (R²) 0.999999
⚠️ Note: These metrics suggest potential data leakage or a deterministic relationship between features and target. Use with caution.
🚀 How to Use
python
from terpredictor import TERModel
model = TERModel.load_pretrained("your-huggingface-username/terpredictor-v1")
input_data = {
"feature_1": 0.12,
"feature_2": 0.03,
...
}
predicted_ter = model.predict(input_data)
⚠️ Limitations
Potential Data Leakage: Extremely high R² may indicate the target is directly derived from input features.
Limited Generalization: Not recommended for predicting TER on unseen or structurally different funds.
No Feature Engineering: Model assumes raw features are sufficient.
📄 License
MIT License
👤 Author
Created by [Your Name or Organization]
📚 Recommendations for Open-Sourcing
Include full training code and preprocessing steps
Provide detailed explanation of evaluation metrics
Add cautionary notes about performance anomalies
Consider publishing a cleaned or anonymized version of the dataset |