- LOADDATA -
In [17]:
pwd
Out[17]:
In [38]:
import pandas as pd
df = pd.read_csv('data.csv')
In [21]:
df.tail()
Out[21]:
- CLEANDATA -
In [39]:
modelRatings = df.pivot_table(index=['EQP_LOCAL_EQP'],columns=['EQP_MODEL_EQP'],values='EQP_FAIL_CNT_EQP').iloc[:, 1:10]
modelRatings.head()
Out[39]:
- SUBRATINGS -
In [25]:
MCARD9060Ratings = modelRatings['MCARD9060']
MCARD9060Ratings.head()
Out[25]:
- PAIRWISE -
In [32]:
similarModels = modelRatings.corrwith(MCARD9060Ratings)
similarModels = similarModels.dropna()
df = pd.DataFrame(similarModels)
df.head(20)
Out[32]:
- SCORE -
In [33]:
similarModels.sort_values(ascending=False).head(20)
Out[33]:
- WRITEUP -
Forecast equipment failure ranked by pairwise correlation using criteria such as model number, subscriber number, and failure count. A dataset of ~400K tuple using model AHTC8717 correlated with model MCARD9060 are the covariants. The algorithm found 6 out of the top 10 equipment failure issues for 60% accuracy