mlnet-predict 1.0.10

MLNetPredict

MLNetPredict is a command-line tool for making predictions using trained ML.NET models. It supports various machine learning scenarios and provides a simple interface for batch predictions.

Features

• Supports multiple ML.NET scenarios:
  • Classification (binary and multi-class)
  • Regression
  • Forecasting
  • Recommendation
  • Text Classification
  • Image Classification
  • Object Detection
• Automatic handling of model dependencies
• Flexible input/output options
• Support for different data formats (CSV, TSV)
• Header detection and delimiter customization

Installation

To install MLNetPredict, you can use the .NET CLI:

dotnet tool install --global mlnet-predict

Usage

Basic syntax:

mlnet-predict <model-path> <input-path> [options]

Required Arguments

• model-path: Path to the directory containing the .mlnet model file
• input-path: Path to the input file or directory (for image-based tasks)

Options

• -o, --output-path: Path to the output file or directory (optional)
• --has-header: Specify if dataset file(s) have header row [true|false]
• --separator: Specify the separator character used in the dataset file(s)

Examples

1. Basic Classification:

mlnet-predict "models/sentiment" "data/input.csv" --has-header true

2. Image Classification with Custom Output:

mlnet-predict "models/image_classifier" "images/test" -o "results/predictions.csv"

3. Forecasting with TSV Input:

mlnet-predict "models/forecast" "data/timeseries.tsv" --separator "\t"

Input Data Format

Text-based Tasks (Classification, Regression, etc.)

• Supported formats: CSV, TSV
• Files should contain the required features as columns
• Headers can be included or excluded (use --has-header option)

Image-based Tasks

• Supported formats: JPG, JPEG, PNG, BMP, GIF
• Input should be a directory containing image files
• Output will be a CSV file with predictions for each image

Output Format

The output format varies depending on the machine learning task:

Classification

PredictedLabel,Score

For multi-class classification:

Top1,Top1Score,Top2,Top2Score,Top3,Top3Score

Regression/Recommendation

Score

Forecasting

PredictedValue,LowerBound,UpperBound

Image Classification

ImagePath,PredictedLabel

Object Detection

ImagePath,PredictedLabels,BoundingBoxes,Scores

Error Handling

The tool provides detailed error messages for common issues:

• Missing or invalid model files
• Unsupported input formats
• Invalid data format
• Missing required columns
• Model loading errors