Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jonatan-leal/ia-proyecto-sustituto/llms.txt

Use this file to discover all available pages before exploring further.

​
Endpoint

POST /train
Trains a Random Forest Classifier model using the data from train.csv and saves it to model.pkl. This endpoint must be called before making predictions.

​
Request

This endpoint does not require any request body or parameters.

​
Headers

No special headers are required for this endpoint.

​
Response

​
message
string
required
A message indicating the training result.
  • Success: "Model successfully trained"
  • Error: "Something went wrong"

​
Training Process

The training endpoint performs the following steps:
  1. Load Training Data: Reads data from train.csv
  2. Encode Categorical Variables:
    • Gender: Female (0), Male (1), Other (2)
    • Smoking History: No Info (0), current (1), ever (2), former (3), never (4), not current (5)
  3. Feature Scaling: Applies StandardScaler to normalize features
  4. Resampling: Uses SMOTEENN to handle class imbalance
  5. Model Training: Trains a Random Forest Classifier
  6. Save Model: Serializes the trained model to model.pkl using pickle
If a model file already exists, it will be overwritten automatically (overwrite mode is enabled by default).

​
Examples

curl -X POST http://localhost/train

​
Response Examples

​
Success Response

Status Code: 200 OK 
{
  "message": "Model successfully trained"
}

​
Error Response

Status Code: 200 OK 
{
  "message": "Something went wrong"
}
Note that even when errors occur during training, the API returns a 200 status code with an error message. Check the response message to determine success or failure.

​
Prerequisites

1

Training Data

Ensure train.csv exists in the API’s working directory with the following columns:
  • gender
  • age
  • hypertension
  • heart_disease
  • smoking_history
  • bmi
  • HbA1c_level
  • blood_glucose_level
  • diabetes (target variable)
2

Dependencies

The following Python packages must be installed:
  • scikit-learn (RandomForestClassifier, StandardScaler)
  • imbalanced-learn (SMOTEENN)
  • pandas
  • pickle
  • loguru
3

Disk Space

Ensure sufficient disk space is available to save the model file (model.pkl).

​
Model Configuration

The trained model uses the following configuration:
​
model
RandomForestClassifier
Random Forest Classifier with default scikit-learn parameters
​
scaler
StandardScaler
StandardScaler for feature normalization
​
resampler
SMOTEENN
SMOTEENN with random_state=42 for handling class imbalance

​
Common Issues

Error: Training fails because train.csv is not found.Solution: Ensure the train.csv file is present in the API’s working directory with the correct column structure.
Error: Training fails due to memory constraints.Solution: Reduce the size of the training dataset or increase available system memory.
Error: Training fails due to incorrect data types in train.csv.Solution: Verify that all columns have the correct data types (strings for categorical, numeric for continuous variables).
Error: Cannot write model file due to permission issues.Solution: Ensure the API has write permissions in the working directory.

​
Training Time

Training time depends on several factors:
  • Dataset size: Larger datasets take longer to train
  • CPU resources: More CPU cores can speed up Random Forest training
  • SMOTEENN resampling: Adds overhead for balancing classes
Typical training time for a dataset with ~100,000 rows: 30-120 seconds

​
Next Steps

After successfully training the model:
  1. Verify that model.pkl was created in the working directory
  2. Use the POST /predict endpoint to make predictions
  3. Monitor model performance and retrain periodically with updated data

Make Predictions

Learn how to use the trained model to make diabetes predictions

Build docs developers (and LLMs) love

Get started for freeTalk to us