Management and Insight Toolkit for Optimization
Start mathematical optimization development today with minto.
pip install minto What is MINTO
Streamlined Experiment Management
Effortlessly manage and track your optimization experiments with minto. Our tool simplifies complex workflows, making it easy to set up, run, and monitor experiments.
Integrated Analytics and Visualization
Gain critical insights with integrated analytics and visualization. minto transforms your data into easy-to-understand visuals, helping you make informed decisions quickly.
Collaboration and Data Sharing (Future)
Future updates to minto will include seamless collaboration and data sharing features. These enhancements will enable teams to work together more effectively, sharing insights and progress in real-time.
Key Features
Easy Logging with Context Management
Begin your experimentation with minto's straightforward logging within a 'with' statement context. This approach allows you to effortlessly capture logs, handling not only scalar values but also more complex data types. When integrated with JijModeling, minto can automatically parse and save the results from optimization solvers, streamlining your data collection process.
import minto
# Create and setup your model and paramters
...
# Run your experiment with minto
experiment = minto.Experiment("Optimization Experiment")
for parameters in parameters_list:
with experiment.run():
experiment.log_parameter("model", problem)
for name, param in parameters.items():
experiment.log_parameter(name, param)
result = solver.solve(problem, **parameters)
Analyzing Experiment Logs with the .table Method
After running experiments, use the .table method to review your collected logs as a pandas DataFrame. This feature is especially handy when working with JijModeling for mathematical modeling and solver results, offering automatic parsing for ease of statistical processing. The auto-parse functionality is designed for compatibility with various mathematical optimization solvers and modelers, further expanding minto's versatility.
# Get your experiment results as a pandas DataFrame
df = experiment.table()
df.head()
Effortless Saving and Sharing of Experimental Results
Easily save your experimental results with minto's save method. Retrieve your data anytime with the .load method, facilitating seamless sharing with other developers. This eliminates the need for designing specific data schemas or writing extensive READMEs for each project, thereby enhancing collaboration and efficiency in your experimentation workflow.
# save your experiment results
experiment.save()
# load your experiment results
experiment = minto.load("Optimization Experiment")
Powered by
