Skip to main content
All CollectionsPredict (Sensai) HelpFeature Guides
Quick Start - Multiple Files, Binary Classification
Quick Start - Multiple Files, Binary Classification
Updated over a week ago

Sensai Customer Journey

Sensai_Customer_Journey.png

Login

You will receive a Welcome email with login instructions.

Click on URL - https://sensai.cloudapps.com

Enter username and temporary password and press the Login button

On your first login, you will be prompted to change your temporary password. Change and press Submit button.

You will then be presented with the Sensai Home Page

Upload Account & Closed Opportunity Files

We need to load some Account and Closed Opportunity information to train our deep learning model.

For this training example you can use the following files - later you will be using your own files or integrating directly with your instance of salesforce.

  1. Download the Academy_Accounts.csv file from here:
    ​https://drive.google.com/file/d/10TI-3JnRdqQWWFXwr9iXu9G3Ygltwcqq/view?usp=sharing

  2. Then Download the Academy_Closed_Opportunities.csv from this location:
    ​https://drive.google.com/file/d/1gVlP4wTo1FVnNtmH8p7IIRWh9NeysqfP/view?usp=sharing
    ​

  3. Press the Options button in the top left of the Navigation Bar and select Files

    mceclip0.png

  4. Press Upload Files button in the top right of the window

  5. Choose the Academy Accounts.csv file and press Upload button. It may take a few seconds to upload the file and for it to appear in the list
    ​

    mceclip1.png

  6. To load the second file, press Upload Files button again

  7. Choose Academy Closed Opportunities.csv and press Upload button. It may take a few seconds to upload the file and for it to appear in the list

Your files are now loaded

Prepare

Create Schema

One of the powerful capabilities of Sensai is the ability to include many data files and then define the relationship between these files and reuse this definition to create different datasets. These can be 1:1 relationships as well as 1:many. At this point we will define a Schema with a many:1 relationship between Opportunity and Account.

  1. Press the Options button in the top left of the Navigation Bar and select Schemas

    mceclip2.png

  2. Press Create Schema button in the top right of the window

  3. Give the Schema a name - e.g. Academy Opportunity with Account

  4. In the Source field, use the dropdown to select 'Other'

    mceclip4.png

  5. Press Add Primary Entity button

  6. Add the primary entity information - in this case the Opportunity

  7. Enter the Primary Key as ID - This is a unique identifier for each Opportunity
    ​

    mceclip8.png

  8. Once this information has been entered, press the Add Entity button

  9. Add the Account entity information and press the Next button

    mceclip9.png

  10. Press the Add Relationship button

  11. Add the relationship information as below and press the Create button

    mceclip10.png

  12. If you drill down into the Schema it should look like this

    mceclip11.png

You have now created the Schema

Create Dataset

  1. Press the Options button in the top left of the Navigation Bar and select Datasets

    mceclip7.png

  2. Press the Create Dataset button in the top right of the window

  3. Give the Dataset a name (Academy Closed Opportunities with Account) and then select your Schema and then choose your files, then press the Create button

    mceclip12.png


    Your Dataset is now created

Create & Train Model

Now it's time to train the model using the Dataset you created in the previous step

  1. Press the Options button in the top left of the Navigation Bar and select Models

  2. Press the Create Model button in the top right of the window

  3. Give the Model a name and choose your Dataset

  4. Select the Binary Classification Architecture Type with your Dataset and Target of ISWON

  5. Then select the Opportunity fields you would like to include in the Model
    ​

    mceclip13.png

  6. Select the Account Tab and then select the Account fields you would like to include in the Model

  7. Then press the Create button

  8. In the Models list, press the dropdown button for the Model you have just created and select Train Model

  9. The deep learning model will be created - this may take some time depending on volume of data and number of fields selected

Predict Open Opportunities & Download

  1. Load Open Opportunity File & Create Dataset

  2. Import the Academy Open Opportunities.csv file - https://drive.google.com/file/d/1OP5skoucWlmcQRtB-5VcvL_ojbw1f1-0/view?usp=sharing

  3. Create a new Dataset “Academy Open Opportunities with Account” from the Academy Open Opportunities.csv and the Academy Accounts.csv files using the Academy Opportunity with Account Schema.

  4. Press the Options button in the top left of the Navigation Bar and select Predictions

  5. Select Request Predictions button

  6. Give the Request a name and choose your Model & Dataset and press the Create button.

    mceclip0.png

This may take a few minutes to create and appear in the list - if it does not appear after a few minutes you may need to refresh the page to see the completed list.

  1. Once completed drill into the Prediction and you will be presented with buttons to Download Prediction and Download Contributions. You will also see 2 charts, the first is a ranked list of the importance of Features and the second is a ranked list of values inside these Features.

  2. Click on the 'Download Prediction' from the dropdown menu button.

This will download the file in a csv format with 3 columns - ID, PROBABILITY and PREDICTION. This can then be imported into either your CRM system or into a BI tool of your choice.

Related Articles
Sensai Releases
Create a Dataset
Define a Single Entity Schema
Quick Start - Single File, Binary Classification
Define a Primary and a Child Entity Schema
Did this answer your question?