Home

TryMe Wiki

Welcome to the TryMe Wiki! This comprehensive resource is dedicated to provide detailed insights into TryMe and its extensive array of features, accompanied by illustrative screenshots. To enhance convenience, we have included a designated folder named "example" within the repository, housing a sample machine learning (ML) model and a set of training and testing CSV data. These files will serve as a reference throughout this wiki, illustrating the practical usage of TryMe. For a concise overview of the tool, we recommend referring to the README file located in the repository.

Background

TryMe was specifically developed to provide business teams and product managers with a straightforward and user-friendly solution for evaluating the effectiveness of machine learning models. It caters to tabular data formats, enabling users to effortlessly modify input values as needed and utilize them for making predictions. The current version of TryMe offers support for XGBoost models, while future updates will introduce compatibility with additional machine learning models.

Prerequisites

• A trained machine learning model
• An input data to be used as example (train.csv or test.csv)

The CSV files inside the "example" folder come from a Kaggle Playground competition whereas the provided XGBoost model is a pre-trained model with some tuning.

Usage

1. Open TryMe on your browser by opening http://localhost:5173/model
2. Upload the XGBoost model by pressing Select Model File button

3. After selecting the file, press Upload button. Upon success, a popup will be shown to the user.

4. Next, upload the example input file by pressing Choose File button under the Upload CSV file header. The preview of the first 5 rows of the file will then be shown to user.

5. Input fields corresponding to the columns on the file will then be shown. Users can choose to copy the values of any of the data rows to the input fields by pressing Copy Value button on the right side of each row. All input fields need to be filled before performing prediction.

6. If users choose training data, which contain the output label, it needs to be removed before performing prediction. This can be done by ticking the selector at the bottom of the page and selecting the output column name in the selector. In the example, train.csv file contains a column called yield, which is the prediction target.

7. After users press on Get Prediction button, a popup dialog will be shown to confirm their selection, alongside with the column they chose to remove, if any.

8. Finally, after pressing OK, the prediction result will be shown