Platform For AI:Predict agricultural loan repayment ability

Dataset fields

Build and run workflow

1. Go to the Machine Learning Designer page.

   1. Log on to the PAI console.

   2. In the left-side navigation pane, click Workspaces. On the Workspaces page, click the name of the workspace that you want to manage.

   3. In the left-side navigation pane, choose Model Training > Visualized Modeling (Designer).

2. Build workflow.

   1. On the Designer page, click Preset Templates.

   2. In Regression Algorithm Implementation for Agricultural Loan Prediction, click Create.

   3. In Create Workflow, configure parameters or keep defaults.

      Workflow Data Storage specifies OSS bucket path for temporary data and models.

   4. Click OK.

      The workflow is created in about 10 seconds.

   5. In the workflow list, select Regression Algorithm for Agricultural Loan Prediction and click Enter Workflow.

   6. Designer automatically builds the workflow.

      

      Area	Description
      ①	Reads workflow datasets: Training: 100 historical loan records with features like farmsize and rainfall to train the model. claimvalue contains the loan amount recovered. Test: 71 current loan applicants. claimvalue contains the requested loan amount. Uses historical data to predict which applicants will receive loan approval.
      ②	Maps string values to numbers. For region, maps north, middle, and south to 0, 1, and 2, then converts to DOUBLE.
      ③	Linear Regression trains and generates a model using historical data. Prediction uses this model to predict loan issuance. Merge Columns combines user ID, predicted value, and requested loan amount. prediction_score contains the predicted repayment amount.
      ④	Regression Model Evaluation evaluates model performance. See Evaluation metrics for details.
      ⑤	Filtering And Mapping identifies eligible applicants whose predicted repayment amount exceeds the requested loan amount.

      Table 1. Evaluation metrics

      Field	Description
      MAE	Mean absolute error.
      MAPE	Mean absolute percentage error.
      MSE	Mean squared error.
      R	Multiple correlation coefficient.
      R2	Determination coefficient.
      RMSE	Root mean squared error.
      SAE	Sum of absolute errors.
      SSE	Sum of squares for error.
      SSR	Sum of squares due to regression.
      SST	Total sum of squares.
      count	Number of rows.
      predictionMean	Mean of predictions.
      yMean	Mean of original dependent variables.

3. Run workflow and view results.

   1. Click above the canvas.

   2. After completion, right-click Filtering and Mapping and choose View Data > Outputs to view eligible applicants.