seq2ftr 0.1.7

A Nice and Convenient Feature Engineering Tool on Sequential Data

This project handles series data

Support Series data Feature Calculation.

Installation

You need Python installed on your system to able to use seq2ftr.

This package contains many feature extraction methods.

Support different type (continues/class) features calculation.

>>> $ pip install seq2ftr

Install Requirements

• numpy

• pandas

• sklearn

Feature Calculation

Support Function

• mean

• max

• min

• freq_of_max

• freq_of_min

• median

• median_mean_distance

• percentage_below_mean

• var

• std

• uniqueCount

• …

Support Type

• 0 - boolean

• 1 - numericla

• 2 - categorical

Example

To start , we load data to python

>>>
import pandas as pd
df = pd.DataFrame([[1,200,"1"],[1,500,"2"],[2,300,"2"],[2,600,"2"]],columns=['id','stock_price',"type"])
df = df.set_index("id")

>>>
from seq2ftr import SequenceTransformer
st_num = SequenceTransformer()
st_num.transformer(df['stock_price']）
# output all features