Merge pull request #55 from mindsdb/staging

Release 24.5.1.0

dataprep_ml/__init__.py

@@ -1,6 +1,6 @@
 from dataprep_ml.base import StatisticalAnalysis, DataAnalysis
 
-__version__ = '0.0.22'
+__version__ = '24.5.1.0'
 __name__ = "dataprep_ml"
 
 

dataprep_ml/cleaners.py

@@ -507,6 +507,10 @@ def clean_timeseries(df: pd.DataFrame, tss: dict) -> pd.DataFrame:
 
     # save original order of columns
     orig_cols = deepcopy(df.columns.to_list())
+
+    # cast order_by as numerical
+    df[tss['order_by']] = pd.to_numeric(df[tss['order_by']], errors='raise')
+
     # fix duplicates by group
     if tss.get('group_by', False):
         correct_dfs = []

dataprep_ml/splitters.py

@@ -57,6 +57,13 @@ def splitter(
     else:
         train, dev, test = simple_split(data, pct_train, pct_dev, pct_test)
 
+    # Final assertions for time series
+    if min(len(train), len(dev)) < tss.get('window', 1):
+        raise Exception(f"Dataset size is too small for the specified window size ({tss.get('window', 1)})")
+
+    if min(len(train), len(dev), len(test)) < tss.get('horizon', 1):
+        raise Exception(f"Dataset size is too small for the specified horizon size ({tss.get('horizon', 1)})")
+
     return {"train": train, "test": test, "dev": dev, "stratified_on": stratify_on}
 
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dataprep-ml"
-version = "0.0.23"
+version = "24.5.1.0"
 description = "Automated dataframe analysis for Machine Learning pipelines."
 authors = ["MindsDB Inc. <[email protected]>"]
 license = "GPL-3.0"