def cleanBill2():
from helper import load_file
from pandas import merge
from CleanComponents import cleanComponents
from sklearn.decomposition import PCA
from pandas import DataFrame
billOfComponents = load_file("bill_of_materials.csv")
billOfComponents['tube_assembly_id'] = billOfComponents.index
components = cleanComponents()
components['component_id'] = components.index
names = components.columns.values
for i in range(1, 9):
cols = names + "_" + str(i)
components.columns = cols
billOfComponents = merge(billOfComponents, components, how='left', on="component_id" + "_" + str(i))
billOfComponents = billOfComponents.drop("component_id" + "_" + str(i), 1)
billOfComponents.index = billOfComponents['tube_assembly_id']
billOfComponents = billOfComponents.drop("tube_assembly_id", 1)
billOfComponents = billOfComponents.fillna(0)
pca = PCA(n_components=20)
pca = pca.fit_transform(billOfComponents)
billOfComponents = DataFrame(pca, billOfComponents.index)
cols = ["pca_" + str(i) for i in billOfComponents.columns.values]
billOfComponents.columns = cols
return billOfComponents
def cleanBill():
from pandas import merge
import numpy as np
from CleanComponents import cleanComponents
components = cleanComponents()
billOfComponents = processBill()
billOfComponents['tube_assembly_id'] = billOfComponents.index
components['component_id'] = components.index
billOfComponents = merge(billOfComponents, components, how='left', on="component_id")
billOfComponents.index = billOfComponents.tube_assembly_id
billOfComponents = billOfComponents.drop('component_id', 1)
billOfComponents = billOfComponents.fillna(0)
quantities = billOfComponents.quantity
comp_numbers = billOfComponents.component_number
ids = billOfComponents.tube_assembly_id
cols = billOfComponents.columns.values.tolist()
cols.remove('quantity')
cols.remove('component_number')
cols.remove('tube_assembly_id')
billOfComponents = billOfComponents[cols].multiply(billOfComponents.quantity, axis="index")
billOfComponents['quantity'] = quantities
billOfComponents['component_number'] = comp_numbers
billOfComponents['tube_assembly_id'] = ids
billOfComponents = billOfComponents.groupby('tube_assembly_id').aggregate(np.sum)
billOfComponents = enrichCompNumber(billOfComponents)
return billOfComponents
