A sklearn-based correlation- and prediction-maker for small csv-data < 10,000 entries. Consquently, no Neural Network will be used and so far the following Models are implemented:
Furthermore, for a first analysis, the cluster- and aprori-pair-plots can be easily generated for checking dependencies in the data.
The CSV-First-Insights-application can be installed like this:
python setup.py install
The options of the Command Line Interface of CSV-First-Insights are:
python -m pyinsights --help
usage: __main__.py [-h] [--fname FNAME FNAME] [--mode MODE] [--export]
Analyzer for small (# < 10,000) csv-Databases with binary content via scikit-learn!
Training-Set and Test-Set is separately stored in two databases.
optional arguments:
-h, --help show this help message and exit
--fname FNAME FNAME Two filenames have to be defined for the train- and test-set.
Default names are: train-data.csv','test-data.csv'
--mode MODE Please chose the model for the forecaset:
*Ridge-Regression as a Variation of Linear-Regressions -> rig(deafault)
*Gradient-Boosting-Trees -> grad
*Random-Forest -> fors
*All three models, please choose -> all
--export Export the Apriori-Analysis, Cluster-Maps, and Predictions as png- and txt-file
The CSV-First-Insights can be also loaded as packages like this:
import pyinsights
import pyinsights.dataread as dr
import pyinsights.mlmodels as ml
import pyinsights.sklsetups as sklThe Ridge-Regression-Prediction of CSV-First-Insights for the The Ultimate Halloween Candy Power Ranking of kaggle:
