def get_diffs(mat):
"""
Applies the given pairwise difference function "pairdiff" to all unique pairs and then returns a list of the results. The length of the list is equal to the length of non NA values choose 2.
"""
diffs = {}
for column in mat:
a = np.array(mat[column], dtype = 'float64') #may need to play with this dtype argument
b = a[~np.isnan(a)]
diffs[column] = [pairdiff(*x) for x in combinations(b,2)]
return diffs
def getPairDiff(
self,
filepath,
querypath,
numberrange,
dayrange,
firstcol,
lastcol,
excel
):
#import data
#determine type of import: csv of excel
filetype = os.path.basename(filepath).split(".")[1]
if filetype == "csv":
data = pd.DataFrame.from_csv(filepath)
elif filetype == "xlsx" or filetype == "xls":
data = pd.read_excel(filepath, sheetname = excel, index_col = 0) #reads the sheet indexed by "excel" - it's python indexed.
self.data = copy(data) #original data
handle_range(data, numberrange, firstcol, lastcol) #process range issue, i.e. [number] to [number]
self.datarange = copy(data) #save the range dataset
#process subset selection from query.xml take data, give datadrop
inclusion = createQuery(querypath, data.shape[0], data)
datadrop = data.ix[[x[0] for x in enumerate(inclusion) if x[1]]] #0 is row, 1 is true if the row should be included
self.dataquery = datadrop #save the query inclusion
#process data with multiple days. 0 is average, 1 is first, 2 is second, -1 is keep all rows #take datadrop, give dataproc
dataproc = handle_days(datadrop, dayrange, firstcol, lastcol)
m, n = dataproc.shape
self.datadays = dataproc
#create difference take dataproc
datamat = dataproc.iloc[:,range(firstcol,lastcol + 1)]
self.datamat = datamat
diff = pairdiff(datamat)
diffmeas = pd.DataFrame(data = diff,
index = dataproc.columns[firstcol:lastcol+1],
columns = ["diff"])
#add summary information to output
outmean, outsd, outnan, outsamplesize = summaries(dataproc, firstcol, lastcol)
diffmeas['mean'] = pd.Series(outmean, index = diffmeas.index)
diffmeas['sd'] = pd.Series(outsd, index = diffmeas.index)
diffmeas['count_nan'] = pd.Series(outnan, index = diffmeas.index)
diffmeas['sample_size'] = pd.Series(outsamplesize, index = diffmeas.index)
self.pairdiff = diffmeas
self.debugger(filepath)
return(diffmeas)
def get_diffs(mat):
"""
Applies the given pairwise difference function "pairdiff" to all unique pairs and then returns a list of the results. The length of the list is equal to the length of non NA values choose 2.
"""
diffs = {}
for column in mat:
a = np.array(
mat[column],
dtype='float64') #may need to play with this dtype argument
b = a[~np.isnan(a)]
diffs[column] = [pairdiff(*x) for x in combinations(b, 2)]
return diffs
def getPairDiff(self, filepath, querypath, numberrange, dayrange, firstcol,
lastcol, excel):
#import data
#determine type of import: csv of excel
filetype = os.path.basename(filepath).split(".")[1]
if filetype == "csv":
data = pd.DataFrame.from_csv(filepath)
elif filetype == "xlsx" or filetype == "xls":
data = pd.read_excel(
filepath, sheetname=excel, index_col=0
) #reads the sheet indexed by "excel" - it's python indexed.
self.data = copy(data) #original data
handle_range(data, numberrange, firstcol,
lastcol) #process range issue, i.e. [number] to [number]
self.datarange = copy(data) #save the range dataset
#process subset selection from query.xml take data, give datadrop
inclusion = createQuery(querypath, data.shape[0], data)
datadrop = data.ix[[
x[0] for x in enumerate(inclusion) if x[1]
]] #0 is row, 1 is true if the row should be included
self.dataquery = datadrop #save the query inclusion
#process data with multiple days. 0 is average, 1 is first, 2 is second, -1 is keep all rows #take datadrop, give dataproc
dataproc = handle_days(datadrop, dayrange, firstcol, lastcol)
m, n = dataproc.shape
self.datadays = dataproc
#create difference take dataproc
datamat = dataproc.iloc[:, range(firstcol, lastcol + 1)]
self.datamat = datamat
diff = pairdiff(datamat)
diffmeas = pd.DataFrame(data=diff,
index=dataproc.columns[firstcol:lastcol + 1],
columns=["diff"])
#add summary information to output
outmean, outsd, outnan, outsamplesize = summaries(
dataproc, firstcol, lastcol)
diffmeas['mean'] = pd.Series(outmean, index=diffmeas.index)
diffmeas['sd'] = pd.Series(outsd, index=diffmeas.index)
diffmeas['count_nan'] = pd.Series(outnan, index=diffmeas.index)
diffmeas['sample_size'] = pd.Series(outsamplesize,
index=diffmeas.index)
self.pairdiff = diffmeas
self.debugger(filepath)
return (diffmeas)
