|
---
|
|
tags:
|
|
- tabular
|
|
- regression
|
|
- tabular-regression
|
|
- dota
|
|
---
|
|
|
|
## Validation Metrics
|
|
|
|
- Accuracy: 0.8284240188362744
|
|
- R2: 0.63
|
|
- MSE: 2428.91
|
|
- MAE: 34.33
|
|
- RMSE: 49.28
|
|
|
|
## Usage
|
|
|
|
```python
|
|
import numpy as np
|
|
from numpy import random
|
|
import pandas as pd
|
|
import onnxruntime as ort
|
|
|
|
# Load the saved file
|
|
model_path = "rd2l_forest.onnx"
|
|
session = ort.InferenceSession(model_path)
|
|
|
|
# Define default naming scheme
|
|
input_name = session.get_inputs()[0].name
|
|
output_name = session.get_outputs()[0].name
|
|
|
|
def prediction(input_data : np.ndarray) -> float
|
|
"""
|
|
Performs inference on the loaded ONNX model using the provided input data.
|
|
|
|
Args:
|
|
input_data (np.ndarray): An array of size (263,), this represents all of a singular players information
|
|
|
|
Returns:
|
|
float: The predicted cost of the player
|
|
|
|
"""
|
|
|
|
# Convert to onnx input format and reshape
|
|
input_data = input_data.to_numpy(dtype=np.float32).reshape(1, -1)
|
|
|
|
# Create prediction
|
|
predictions = session.run([output_name], {input_name: input_data})
|
|
|
|
# Convert to individual value
|
|
return round(float(predictions[0][0][0]), 2)
|
|
|
|
sample_df = pd.DataFrame(np.random.rand(263))
|
|
|
|
prediction(sample_df)
|
|
```
|
|
---
|
|
license: mit
|
|
---
|
|
|
|
|
