Automated validation of A.I.
With VERIFY AI, you can easily validate the quality of the training dataset and the performance of the trained artificial intelligence model through a report.
With VERIFY AI, you can easily validate the quality of the training dataset and the performance of the trained artificial intelligence model through a report.
VERIFY AI automatically generates performance reports after data and model validation and provides continuous quality control.
| Index | Column name | Training data usage status | Data | Number | Loss |
|---|---|---|---|---|---|
| 1 | Warehouse block | True | verify_data.csv | 9899 | 0 |
| 2 | Delivery method | True | verify_data.csv | 9899 | 0 |
| 3 | Customer review | True | verify_data.csv | 9899 | 0 |
| 4 | Previous purchase | True | verify_data.csv | 9899 | 0 |
| Index | Column name | unique key | Type | Minimum value | Maximum value |
|---|---|---|---|---|---|
| 1 | Warehouse block | 5 | object | - | - |
| 2 | Delivery method | 3 | object | - | - |
| 3 | Customer review | 5 | number | 1.0 | 5.0 |
| 4 | Previous purchase | 8 | number | 2.0 | 10.0 |
| Index | Column name | Standard value | Average value | Highest rank | Frequency |
|---|---|---|---|---|---|
| 1 | Warehouse block | - | - | 0.0 | 3305.0 |
| 2 | Delivery method | - | - | 0.0 | 6723.0 |
| 3 | Customer review | 1.411328 | 2.991514 | - | - |
| 4 | Previous purchase | 1.515191 | 3.565815 | - | - |
Upload the source dataset to view the statistical analysis results immediately.
You can verify the bias and the statistical validity of prepared training datasets, such as duplicates and blanks.
Automatically train more than 100 models and compare and validate them with your own models.
Statistical analysis and verification results for data and models can be automatically generated and printed as reports.
| Index | Column name | Training data usage status | Data | Numbers | Loss |
|---|---|---|---|---|---|
| 1 | Warehouse block | True | verify_data.csv | 338,000 | 0 |
| 2 | Delivery method | True | verify_data.csv | 338,000 | 0 |
| 3 | Customer review | True | verify_data.csv | 338,000 | 0 |
| 4 | Previous purchase | True | verify_data.csv | 338,000 | 0 |
| Index | Column name | unique key | Type | Minimum value | Maximum value |
|---|---|---|---|---|---|
| 1 | Warehouse block | 5 | object | - | - |
| 2 | Delivery method | 3 | object | - | - |
| 3 | Customer review | 5 | number | 1.0 | 5.0 |
| 4 | Previous purchase | 8 | number | 2.0 | 10.0 |
You can create statistics by simply uploading the datasets to create a data preprocessing strategy. You can also review and ensure the reliability of your final training dataset.
Automatic visualization allows you to view large datasets at a glance, and the visualized metrics of the generated verification model allow you to evaluate without any analysis.
| Overall_Statistics | Value |
|---|---|
| SOA1(Landis & Koch) | Substantial |
| SOA2(Fleiss) | Excellent |
| SOA3(Altman) | Good |
| SOA4(Cicchetti) | Excellent |
| SOA6(Matthews) | Strong |
| Overall ACC | 0.8 |
| Kappa | 0.7802197802197802 |
| Overall RACC | 0.09 |
| ACC Macro | 0.9714285714285713 |
| F1 Macro | 0.7040816326530612 |
For the training dataset, you can verify up to 100 AI models that can be derived based on DS2.ai's AutoML engine, and review the validity of the training dataset itself based on model indicators.