Predict student grades using the built-in logistic regression algorithm template in Designer-Platform For AI(PAI)-阿里云帮助中心

Designer provides a preset logistic regression template to quickly build a model that predicts final exam scores based on student family backgrounds and school behaviors. This helps you identify the key factors that affect academic performance. This topic describes how to use the preset logistic regression template.

Background

After you obtain the student score prediction model from this pipeline, you can upload data for prediction to a MaxCompute table to perform offline predictions.

Prerequisites

You have created a workspace. For more information, see Create and manage workspaces.
You have associated MaxCompute resources with your workspace. For more information, see Create and manage workspaces.

Dataset

The dataset for this pipeline consists of 25 feature columns and one target column. The following table describes the fields.

Parameter	Type	Description
sex	STRING	Gender. F indicates female, and M indicates male.
address	STRING	Address. U indicates city, and R indicates rural.
famsize	STRING	Number of family members. LE3 indicates three or fewer people, and GT3 indicates more than three people.
pstatus	STRING	Specifies whether you live with your parents. T indicates living with your parents, and A indicates living apart from your parents.
medu	DOUBLE	The mother's education level, on a scale of 0 to 4. A higher value indicates a higher education level.
fedu	DOUBLE	The father's education level, on a scale of 0 to 4. A higher value indicates a higher education level.
mjob	STRING	The mother's job, including jobs in education, healthcare, and the service industry.
fjob	STRING	The father's job, including jobs in education, healthcare, and the service industry.
guardian	STRING	The student's guardian. Valid values are `mother`, `father`, and `other`.
traveltime	DOUBLE	The travel time from home to school, in minutes.
studytime	DOUBLE	The weekly study time, in hours.
failures	DOUBLE	The number of past class failures.
schoolsup	STRING	Specifies whether additional learning assistance is available. The value can be yes or no.
fumsup	STRING	Specifies whether a private tutor is hired. The value can be yes or no.
paid	STRING	Specifies whether assistance is provided for related exam subjects. The value can be yes or no.
activities	STRING	Indicates whether the user is enrolled in extracurricular classes. The value can be yes or no.
higher	STRING	Specifies whether to pursue further education. The value is yes or no.
internet	STRING	Specifies whether the home is connected to the internet. The value can be yes or no.
famrel	DOUBLE	The quality of family relationships, on a scale of 1 (very bad) to 5 (excellent).
freetime	DOUBLE	The amount of free time after school, on a scale of 1 (very low) to 5 (very high).
goout	DOUBLE	The frequency of going out with friends, on a scale of 1 (very low) to 5 (very high).
dalc	DOUBLE	Workday alcohol consumption, on a scale of 1 (very low) to 5 (very high).
walc	DOUBLE	Weekend alcohol consumption, on a scale of 1 (very low) to 5 (very high).
health	DOUBLE	The current health status, on a scale of 1 (very bad) to 5 (very good).
absences	DOUBLE	The number of school absences, from 0 to 93.
g3	STRING	The final exam score, on a 20-point scale.

The following figure shows a sample of the pipeline data. 实验示例数据

Procedure

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).

Build the pipeline.

On the Designer page, click the Preset Templates tab.
In the template list, find the Online Prediction - Student Score Prediction section and click Create.
In the Create Pipeline dialog box, configure the parameters. You can use the default settings for all parameters.

The Data Storage parameter specifies an OSS Bucket path to store temporary data and models generated during the pipeline run.
Click Confirm.

It takes about 10 seconds to create the pipeline.
On the Pipelines tab, double-click the Online Prediction - Student Score Prediction pipeline to open it.

Designer automatically builds a pipeline based on the preset template, as shown in the following figure.

预测成绩实验

Area	Description
①	Data preprocessing. The SQL Script component is used to structure the text data: Convert yes and no in the source data to 0 and 1, respectively. For text fields that contain multiple categories, abstract the data based on your business scenario. For example, for the Mjob field, you can represent the value teacher as 1 and all other values as 0. In this case, the abstracted feature indicates whether the job is related to education. For the target column, set values greater than 18 to 1 and all other values to 0.
②	The Normalize component performs normalization to scale all fields to a range of 0 to 1. This eliminates the impact of differing value scales.
③	The input dataset is split into a training dataset and a prediction dataset at an 8:2 ratio.
④	This step uses a logistic regression algorithm to train and generate an offline model.
⑤	The Confusion Matrix component evaluates the model's accuracy.

Run the pipeline and view the results.
1. Above the canvas, click the Run icon to run the pipeline.
2. After the pipeline finishes running, right-click the Confusion Matrix component on the canvas and click Visual Analysis in the shortcut menu.
3. In the Confusion Matrix dialog box, click the Summary tab to view the model's prediction accuracy, which is over 80%.

Predict student exam scores

Background

Prerequisites

Dataset

Procedure

Related documents