def getCategoryDistributionInRegions(dcPaths, dfCellLocType):
'''
This function computes:
1. total user number in different types of locations;
2. category user number in different types of locations;
3. category traffic in different types of locations;
param:
dcPaths - roaming paths
dfCellLocType - index=lac-cid
return:
srUserNumPerRegion - total user number per region, index:region
dfCategoryUserNumPerRegion - category user number per region, row:category, col:region
dfCategoryTrafficPerRegion - category traffic per region, row:category, col:region
'''
dcTotalUserNumPerRegion = {} # total user number in different types of locations;
dcCategoryUserNumPerRegion = {} # category user number in different types of locations;
dcCategoryTrafficPerRegion = {} # category traffic in different types of locations;
for path in dcPaths.values():
for node in path.m_lsNodes:
strKey = "%d-%d" % (node.m_nLac, node.m_nCellID)
nRegionType = dfCellLocType.loc[strKey]["typeID"]
strRegionName = region_type.g_dcRegionTypeName[nRegionType]
# total user
updateDictBySum(dcTotalUserNumPerRegion, strRegionName, 1)
dcCategoryUserNumForCurrentRegion = dcCategoryUserNumPerRegion.get(strRegionName, None)
if (dcCategoryUserNumForCurrentRegion is None):
dcCategoryUserNumForCurrentRegion = {}
dcCategoryUserNumPerRegion[strRegionName] = dcCategoryUserNumForCurrentRegion
dcCategoryTrafficForCurrentRegion = dcCategoryTrafficPerRegion.get(strRegionName, None)
if (dcCategoryTrafficForCurrentRegion is None):
dcCategoryTrafficForCurrentRegion = {}
dcCategoryTrafficPerRegion[strRegionName] = dcCategoryTrafficForCurrentRegion
dcCategoryUserNumForCurretNode = {}
for app in node.m_lsApps:
strCategoryName = app_category.getAppCategory(app.m_nServiceType)
dcCategoryUserNumForCurretNode[strCategoryName] = 1
updateDictBySum(dcCategoryTrafficForCurrentRegion, strCategoryName, app.m_nDownBytes)
for (k,v) in dcCategoryUserNumForCurretNode.iteritems():
updateDictBySum(dcCategoryUserNumForCurrentRegion, k, v)
srUserNumPerRegion = pd.Series(dcTotalUserNumPerRegion)
dfCategoryUserNumPerRegion = pd.DataFrame(dcCategoryUserNumPerRegion)
dfCategoryTrafficPerRegion = pd.DataFrame(dcCategoryTrafficPerRegion)
# delete unknown region type & app cateogry
srUserNumPerRegion.drop(labels=region_type.g_dcRegionTypeName[region_type.ID_TYPE_UNKNOWN], inplace=True)
dfCategoryUserNumPerRegion.drop(labels=app_category.g_strUnknown, inplace=True)
del dfCategoryUserNumPerRegion[region_type.g_dcRegionTypeName[region_type.ID_TYPE_UNKNOWN]]
dfCategoryTrafficPerRegion.drop(labels=app_category.g_strUnknown, inplace=True)
del dfCategoryTrafficPerRegion[region_type.g_dcRegionTypeName[region_type.ID_TYPE_UNKNOWN]]
return srUserNumPerRegion, dfCategoryUserNumPerRegion, dfCategoryTrafficPerRegion
