search

dotnet / machinelearning-samples

Samples for ML.NET, an open source and cross-platform machine learning framework for .NET.
https://dot.net/ml
MIT License
4.42k stars 2.66k forks source link

Error mlcli - when training regression test. #652

Open julianobd opened 4 years ago

julianobd commented 4 years ago

mlnet auto-train --task regression --dataset taxi-fare-train.csv --test-dataset taxi-fare-test.csv --label-column-name fare_amount --max-exploration-time 1000

return:

`Exploring multiple ML algorithms and settings to find you the best model for ML task: regression For further learning check: https://aka.ms/mlnet-cli

▓▓▓▓▓▓▓▓▓▓▓───────────────────────────────────────────────────────────────────────────────────────────────────────────── Best quality(RSquared): 0,9440, Best Algorithm: LightGbmRegression, Last Algorithm: FastTreeTweedieRegression
00:01:37Exception occured while exploring pipelines: All instances skipped due to missing features. Please see the log file for more info.`

Log: debug_log.txt

Datasets: teste.zip

CESARDELATORRE commented 4 years ago

Can you try with a single input dataset (so the CLI will internally use cross-validation instead of provided split files) just to isolate if the issue is with the test-dataset?

julianobd commented 4 years ago

I've just tryied using only train dataset and only test dataset.

test_dataset_debug_log.txt

train_dataset_debug_log.txt

commands used: mlnet auto-train --task regression --dataset taxi-fare-test.csv --label-column-name fare_amount --max-exploration-time 1000

mlnet auto-train --task regression --dataset taxi-fare-train.csv --label-column-name fare_amount --max-exploration-time 1000