How to Use TI-ONE Built‑in Operators for the 2020 Tencent Advertising Algorithm Competition

1. Create Project and Workflow

Log in to the TI‑ONE console, switch the region to Shanghai, then on the project list click My Projects → New Project . Fill in the project information and select a COS bucket (Shanghai region) to store training data and intermediate results.

After the project is created, click the "+" button inside the project to create a custom workflow and open the canvas.

2. Data Ingestion

In the canvas, drag Input → Public Dataset → Algorithm Competition Dataset onto the canvas. This dataset contains both training and test sets, completing the data ingestion step.

3. Model Training

3.1 Choose Operator

Select the Decision Tree Classification operator from the left‑hand algorithm palette and drag it onto the canvas. Connect the training‑set output pin to the operator’s input pin, and connect the test‑set output pin to the model output pin (the small beaker icon). The platform automatically generates the data I/O paths.

3.2 Configure Algorithm Parameters

Click the operator to open the right‑hand dialog and set the algorithm I/O parameters as follows:

Input file type: csv

Feature columns: 0‑3

Label column: 4

Header included: yes

Delimiter: comma

Use the default algorithm parameters for this example.

3.3 Configure Model Parameters

Click the small beaker (model) on the left side of the operator to open the model configuration dialog.

Model update mode: manual

Model run mode: automatic

Model I/O parameters:

Input/Output file type: csv

Feature columns: 0‑3

Label column: 4

Header included: yes

Delimiter: comma

4. Run Workflow

After all configurations are complete, click Save at the top of the canvas, then click Run to execute the workflow.

5. Result Upload

5.1 Download Result Files

Right‑click the model (small beaker), choose Model Operation → View Data , then click Go to COS to view in the popup. Locate the result file in the COS bucket and download it.

5.2 Process and Upload Result Files

Format the result file according to the competition requirements, then upload the final file to the designated COS bucket.

5.3 Get Result URL

In this example, files starting with 129 are the submission results. Click Details to obtain the Object URL , which can be used for the official submission.

With these steps, you have completed the end‑to‑end process of using TI‑ONE built‑in operators to train a model and submit results for the competition.