![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4172517%2Fd7f40b50c1ff56253e17da4e2fdbfdda%2FIMG_20231127_114450_1.jpg?generation=1701075336921467&alt=media\")
Don't forget to move the peaks.
\nFor values that are greater than the maximum of Train column (0):
\nY = Y + (Y- max(Ycol0))*0.28
For min:
\nY = Y + (Y- min(Ycol0))*0.28
\nRepeat the process for every column between 0 and 18116.
code : https://www.kaggle.com/code/liudacheldieva/submit-only
\nversion 11
pandas pd\nfn = \ndf57d = pd.read_csv(, index_col=)\n\ndf_de_train = pd.read_parquet(fn)\naggs= df_de_train[ (df_de_train.columns)[:]].T[].copy()\naggs = pd.DataFrame(aggs)\ndiff_= pd.DataFrame((df_de_train[ (df_de_train.columns)[:]] .(axis=) - df_de_train[ (df_de_train.columns)[:]] .(axis=)))\nmin_= pd.DataFrame(df_de_train[ (df_de_train.columns)[:]] .(axis=) )\nmax_= pd.DataFrame(df_de_train[ (df_de_train.columns)[:]] .(axis=) )\naggs[]=diff_[]\naggs[]=min_[]\naggs[]=max_[] \n\n col (df57d.columns):\n df[df[col] > aggs.loc[col] [] ] = df[df[col] > aggs.loc[col] [] ] + (df[df[col] > aggs.loc[col] [] ]- aggs.loc[col] [])/*\n df[df[col] < aggs.loc[col] [] ] = df[df[col] < aggs.loc[col] [] ] - ( aggs.loc[col] [] -df[df[col] < aggs.loc[col] [] ])/*\nDon't forget to move the peaks.
\nFor values that are greater than the maximum of Train column (0):
\nY = Y + (Y- max(Ycol0))*0.28
For min:
\nY = Y + (Y- min(Ycol0))*0.28
\nRepeat the process for every column between 0 and 18116.
code : https://www.kaggle.com/code/liudacheldieva/submit-only
\nversion 11
pandas pd\nfn = \ndf57d = pd.read_csv(, index_col=)\n\ndf_de_train = pd.read_parquet(fn)\naggs= df_de_train[ (df_de_train.columns)[:]].T[].copy()\naggs = pd.DataFrame(aggs)\ndiff_= pd.DataFrame((df_de_train[ (df_de_train.columns)[:]] .(axis=) - df_de_train[ (df_de_train.columns)[:]] .(axis=)))\nmin_= pd.DataFrame(df_de_train[ (df_de_train.columns)[:]] .(axis=) )\nmax_= pd.DataFrame(df_de_train[ (df_de_train.columns)[:]] .(axis=) )\naggs[]=diff_[]\naggs[]=min_[]\naggs[]=max_[] \n\n col (df57d.columns):\n df[df[col] > aggs.loc[col] [] ] = df[df[col] > aggs.loc[col] [] ] + (df[df[col] > aggs.loc[col] [] ]- aggs.loc[col] [])/*\n df[df[col] < aggs.loc[col] [] ] = df[df[col] < aggs.loc[col] [] ] - ( aggs.loc[col] [] -df[df[col] < aggs.loc[col] [] ])/*\nIt is possible to raise or lower the model for every drug.
\nTo determine the distance to move, examine \"minimizing drug RMSE\"'.
Create a single model for predicting the \"single variable X\" of the drug.
\n distance = single_variable_X \n - X_minimized_drug_RMSE\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4172517%2Fbe118dda0e76d9807f7dd5dc440a577b%2FIMG_20231127_141llll225.jpg?generation=1701084758977754&alt=media\")
You can calculate min and max values column wise and then multiply them by 0.72 and use numpy's clip function finally.
", "rawMarkdown": "You can calculate min and max values column wise and then multiply them by 0.72 and use numpy's clip function finally.", "votes": null } ], "comments": [ { "id": 2539882, "author_name": "", "author_url": "", "post_date": "11/27/2023 11:30:46", "content": "It is possible to raise or lower the model for every drug.
\nTo determine the distance to move, examine \"minimizing drug RMSE\"'.
Create a single model for predicting the \"single variable X\" of the drug.
\n distance = single_variable_X \n - X_minimized_drug_RMSE\n
You can calculate min and max values column wise and then multiply them by 0.72 and use numpy's clip function finally.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2539662": "Don't forget to move the peaks.\n\n\n\nFor values that are greater than the maximum of Train column (0):\nY = Y + (Y- max(Ycol0))*0.28 \n\nFor min:\nY = Y + (Y- min(Ycol0))*0.28\nRepeat the process for every column between 0 and 18116.\n\n\n\ncode : https://www.kaggle.com/code/liudacheldieva/submit-only\nversion 11\n\n```python\nimport pandas as pd\nfn = '/kaggle/input/open-problems-single-cell-perturbations/de_train.parquet'\ndf57d = pd.read_csv('/kaggle/input/op2-gentle-param-tuner/submission_tsvd50_Ridge_LeaveOneOutEncoderCompoundCellType_blendWithPriors_0.45_0.1MT_tuned.csv', index_col='id')\n\ndf_de_train = pd.read_parquet(fn)# , index_col = 0)\naggs= df_de_train[ list(df_de_train.columns)[5:]].T[0].copy()\naggs = pd.DataFrame(aggs)\ndiff_= pd.DataFrame((df_de_train[ list(df_de_train.columns)[5:]] .max(axis=0) - df_de_train[ list(df_de_train.columns)[5:]] .min(axis=0)))\nmin_= pd.DataFrame(df_de_train[ list(df_de_train.columns)[5:]] .min(axis=0) )\nmax_= pd.DataFrame(df_de_train[ list(df_de_train.columns)[5:]] .max(axis=0) )\naggs['diff']=diff_[0]\naggs['min']=min_[0]\naggs['max']=max_[0] \n# second moov22\nfor col in list(df57d.columns):\n df[df[col] > aggs.loc[col] ['max'] ] = df[df[col] > aggs.loc[col] ['max'] ] + (df[df[col] > aggs.loc[col] ['max'] ]- aggs.loc[col] ['max'])/100*0.18# 08\n df[df[col] < aggs.loc[col] ['min'] ] = df[df[col] < aggs.loc[col] ['min'] ] - abs ( aggs.loc[col] ['min'] -df[df[col] < aggs.loc[col] ['min'] ])/100*0.18# 08\n\n```", "2539882": "It is possible to raise or lower the model for every drug.\nTo determine the distance to move, examine \"minimizing drug RMSE\"'.\n\nCreate a single model for predicting the \"single variable X\" of the drug.\n\n```python\n distance = single_variable_X \n - X_minimized_drug_RMSE\n```\n", "2543705": "You can calculate min and max values column wise and then multiply them by 0.72 and use numpy's clip function finally." }, "source": "meta" }