Hello,
I am working on creating a LightGBM model using a JMP Python script (JMP18).
My current setup is as follows:
- X variables: "train1.csv" file, columns D1 through D1776
- Y variable: "y1.csv" file, "Activity" column
- Prediction target "Activity": "test1.csv" file, columns D1 through D1776
- Note: Original file "train_original.csv" file, columns Activity through D1776. from Kaggle.com (Predicting a Biological Response)
I would appreciate it if somebody can provide the guidance for the following issues:
- Cannot perform LightGBM modeling in JMP18 python environment
I've included my JMP Python script below for your reference. See the followings for details.
Thank you for your time in advance! : )
import jmp
import jmputils
# update to latest version of pip (Package Installer of python) and setuptools then install numpy & pandas
jmputils.jpip('install --upgrade', 'pip setuptools')
jmputils.jpip('install', 'pandas numpy scikit-learn keras lightgbm')
# Import package
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
# Load data
train1 = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/train1.csv')
y1 = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/y1.csv')
test1 = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/test1.csv')
train1.head() # No Result in JMP, but it's okay. It's fine to open the datatable using 'jmp.open'
y1.head() # No Result in JMP, but it's okay. It's fine to open the datatable using 'jmp.open'
# train1 = jmp.open('D:/steve.kim/Kaggle/BI Biological Response/train1.csv')
# y1 = jmp.open('D:/steve.kim/Kaggle/BI Biological Response/y1.csv')
# 01 Modeling training - Library import
from lightgbm import LGBMClassifier
# 02 Modeling training - LGBM Baseline model without hyperparameter tuning
lgb = LGBMClassifier()
# 03 Modeling training - define X (factors) and Y (responses) variables
lgb.fit(train1, y1) # lightgbm.basic.LightGBMError: Length of labels differs from the length of #data
# 04 Predict
predslgb = lgb.predict_proba(test1)
Sorry about this question!
The python pandas and lightGBM are working well in JMP18 python script! : )
import jmp
import jmputils
# update to latest version of pip (Package Installer of python) and setuptools then install numpy & pandas
jmputils.jpip('install --upgrade', 'pip setuptools')
jmputils.jpip('install', 'pandas numpy scikit-learn keras lightgbm')
# Checking package version
jmputils.jpip('list')
# Import package
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
# Any results you write to the current directory are saved as output.
import os
print (os.listdir("D:/steve.kim/Kaggle/BI Biological Response"))
# Load data
train_x = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/train_x.csv')
print(train_x.head())
train_y = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/train_y.csv')
print(train_y.head())
test_x = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/test_x.csv')
print(test_x)
# Modeling
from lightgbm import LGBMClassifier
lgb = LGBMClassifier(colsample_bytree=0.6, subsample=0.8)
lgb.fit(train_x, train_y)
preds_lgb = lgb.predict_proba(test_x)
sub_lgb = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/test_y.csv')
sub_lgb["Activity"] = preds_lgb[:,1]
print(sub_lgb.head())
test_answer = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/test_answer.csv')
# Evaluation
from sklearn.metrics import mean_absolute_error, mean_squared_error, mean_absolute_percentage_error
print('LightGBM MAE')
print(mean_absolute_error(test_answer, sub_lgb))
print('LightGBM MSE')
print(mean_squared_error(test_answer, sub_lgb))Re: How can I perform Machine Learning Modeling (LGBM) using JMP18 Python Script?
Oh... The code 'head()' is woking well after I consider it as a terminal environment not the jupyter notebook!
y1 = pd.read_csv('D:/steve.kim/Kaggle/BI Biological Response/y1.csv')
print(y1.head()) then I can get the code result from embedded Log!
So I think I can resolve the rest of things as well.
Activity
0 1
1 1
2 1
3 1
4 1
