Privacy computing fat - offline prediction

One 、 explain

Fate The model predictions are offline forecast and Online forecasting Two ways , The effect is the same , Mainly the way of use 、 Applicable scenario 、 High availability 、 There are great differences in performance, etc ; This article uses Fate be based on Vertical logistic regression The model trained by the algorithm is used for off-line prediction practice .

• Based on the above 《 Privacy computing FATE - model training 》 The model trained in the
• About Fate Please refer to the article for the basic overview and installation deployment of 《 Privacy computing FATE - Key concepts and stand-alone Deployment Guide 》

 

Two 、 Query model information

Execute the following command , Get into Fate In the container ：

docker exec -it $(docker ps -aqf "name=standalone_fate") bash

First, we need to get the corresponding model_id and model_version Information , Can pass job_id Execute the following command to get ：

flow job config -j 202205070226373055640 -r guest -p 9999 --output-path /data/projects/fate/examples/my_test/

job_id Can be in FATE Board View in .

After successful execution, the corresponding model information will be returned , And generate a folder under the specified directory job_202205070226373055640_config

{
    "data": {
        "job_id": "202205070226373055640",
        "model_info": {
            "model_id": "arbiter-10000#guest-9999#host-10000#model",
            "model_version": "202205070226373055640"
        },
        "train_runtime_conf": {}
    },
    "retcode": 0,
    "retmsg": "download successfully, please check /data/projects/fate/examples/my_test/job_202205070226373055640_config directory",
    "directory": "/data/projects/fate/examples/my_test/job_202205070226373055640_config"
}

job_202205070226373055640_config It contains 4 File ：

• dsl.json： Mission dsl To configure .
• model_info.json： Model information .
• runtime_conf.json： Task running configuration .
• train_runtime_conf.json： empty .

 

3、 ... and 、 Model deployment

Execute the following command ：

flow model deploy --model-id arbiter-10000#guest-9999#host-10000#model --model-version 202205070226373055640

Pass respectively --model-id And --model-version Specify the... Found in the above steps model_id and model_version

After the deployment is successful, return ：

{
    "data": {
        "arbiter": {
            "10000": 0
        },
        "detail": {
            "arbiter": {
                "10000": {
                    "retcode": 0,
                    "retmsg": "deploy model of role arbiter 10000 success"
                }
            },
            "guest": {
                "9999": {
                    "retcode": 0,
                    "retmsg": "deploy model of role guest 9999 success"
                }
            },
            "host": {
                "10000": {
                    "retcode": 0,
                    "retmsg": "deploy model of role host 10000 success"
                }
            }
        },
        "guest": {
            "9999": 0
        },
        "host": {
            "10000": 0
        },
        "model_id": "arbiter-10000#guest-9999#host-10000#model",
        "model_version": "202205070730131040240"
    },
    "retcode": 0,
    "retmsg": "success"
}

After successful deployment, a new model_version

 

Four 、 Prepare forecast configuration

Execute the following command ：

cp /data/projects/fate/examples/dsl/v2/hetero_logistic_regression/hetero_lr_normal_predict_conf.json /data/projects/fate/examples/my_test/

Put... Directly Fate The built-in vertical logistic regression algorithm prediction configuration example , Copy it to our my_test Under the table of contents .

The predicted configuration file mainly configures three parts ：

• The above section is to configure the initiator and participant roles
• The middle part needs to be filled with the correct Model information
• The following is the data table used in the forecast

The only thing that needs to be modified is the middle one Model information part ; Note that the version number entered here is Model deployment Version number returned after , And need to add job_type by predict Specify the task type as forecast task .

 

5、 ... and 、 Perform forecast task

Execute the following command ：

flow job submit -c hetero_lr_normal_predict_conf.json

Just like model training, it also uses submit command , adopt -c Specify profile .

Return after successful execution ：

{
    "data": {
        "board_url": "http://127.0.0.1:8080/index.html#/dashboard?job_id=202205070731385067720&role=guest&party_id=9999",
        "code": 0,
        "dsl_path": "/data/projects/fate/fateflow/jobs/202205070731385067720/job_dsl.json",
        "job_id": "202205070731385067720",
        "logs_directory": "/data/projects/fate/fateflow/logs/202205070731385067720",
        "message": "success",
        "model_info": {
            "model_id": "arbiter-10000#guest-9999#host-10000#model",
            "model_version": "202205070730131040240"
        },
        "pipeline_dsl_path": "/data/projects/fate/fateflow/jobs/202205070731385067720/pipeline_dsl.json",
        "runtime_conf_on_party_path": "/data/projects/fate/fateflow/jobs/202205070731385067720/guest/9999/job_runtime_on_party_conf.json",
        "runtime_conf_path": "/data/projects/fate/fateflow/jobs/202205070731385067720/job_runtime_conf.json",
        "train_runtime_conf_path": "/data/projects/fate/fateflow/jobs/202205070731385067720/train_runtime_conf.json"
    },
    "jobId": "202205070731385067720",
    "retcode": 0,
    "retmsg": "success"
}

 

6、 ... and 、 View forecast results

Can be returned by board_url perhaps job_id Go to FATE Board View results in , But the graphical interface can only be viewed at most 100 Bar record ;

We can go through output-data command , Export all data output of the specified component ：

flow tracking output-data -j 202205070731385067720 -r guest -p 9999 -cpn hetero_lr_0 -o /data/projects/fate/examples/my_test/predict

• -j： Specify the... Of the forecast task job_id
• -cpn： Specify the component name .
• -o： Specify the output directory .

Return after successful execution ：

{
    "retcode": 0,
    "directory": "/data/projects/fate/examples/my_test/predict/job_202205070731385067720_hetero_lr_0_guest_9999_output_data",
    "retmsg": "Download successfully, please check /data/projects/fate/examples/my_test/predict/job_202205070731385067720_hetero_lr_0_guest_9999_output_data directory"
}

In the catalog /data/projects/fate/examples/my_test/predict/job_202205070731385067720_hetero_lr_0_guest_9999_output_data You can see two files in ：

• data.csv： For all data output .
• data.meta： Is the column header of the data .

 

Code scanning, attention, surprise ！