#!/usr/bin/env python
# coding: utf-8

# # NumPy Data Access Using ArcPy 

# In[1]:


import arcpy as ARCPY
import arcpy.da as DA
inputFC = r'../data/CA_Polygons.shp'
fieldNames = ['PCR2000', 'POP2000', 'PERCNOHS']
tab = DA.TableToNumPyArray(inputFC, fieldNames)
print(tab)


# # SSDataObject
# 1. Environment Settings (Except Extent)
# 2. Bad Records
# 3. Error/Warning Messages
# 4. Localization
# 5. **Feature Accounting**
#   * Cursors and DataAccess are not assured to read attributes in order.
# 
#   * Keeps track of the shapes and their attributes so that one can create output features w/o post-joins.
# 
#   * Unique ID works with Spatial Weights Formats in ArcGIS, PySAL, R, Matlab, GeoDa etc..

# In[38]:


import SSDataObject as SSDO
import os as OS
inputFC = r'../data/CA_Polygons.shp'
fullFC = OS.path.abspath(inputFC)
fullPath, fcName = OS.path.split(fullFC)
ssdo = SSDO.SSDataObject(inputFC)
uniqueIDField = "MYID"
fieldNames = ['PCR2010', 'POP2010', 'PERCNOHS']
ssdo.obtainData(uniqueIDField, fieldNames)
ssdo = SSDO.SSDataObject(inputFC)
ssdo.obtainData("MYID", fieldNames)
print(ssdo.fields['POP2010'].data)


# # Using PANDAS to get that R Feel

# In[39]:


import pandas as PANDAS
df = ssdo.getDataFrame()
print(df)


# # Advanced Analysis [SciPy Example - KMeans]

# In[40]:


import numpy as NUM
import scipy.cluster.vq as CLUST
import arcgisscripting as ARC
X = df.as_matrix()
whiteData = CLUST.whiten(X)
centers, distortion = CLUST.kmeans(whiteData, 6)
groups = ARC._ss.closest_centroid(whiteData, centers)
print(groups)


# # Max-P Regions Using PySAL

# In[41]:


import pysal as PYSAL
import pysal2ArcGIS as PYSAL_UTILS
swmFile = OS.path.join(fullPath, "rook_bin.swm")
w = PYSAL_UTILS.swm2Weights(ssdo, swmFile)
maxp = PYSAL.region.Maxp(w, X[:,0:2], 3000000., floor_variable = X[:,2])
maxpGroups = NUM.empty((ssdo.numObs,), int)
for regionID, orderIDs in enumerate(maxp.regions):
    maxpGroups[orderIDs] = regionID
maxpGroups


# # SKATER for Comparison

# In[42]:


import Partition as PART
skater = PART.Partition(ssdo, fieldNames, spaceConcept = "GET_SPATIAL_WEIGHTS_FROM_FILE",
                        weightsFile = swmFile, kPartitions = 6)
print(skater.partition)


# In[43]:


ARCPY.env.overwriteOutput = True
outputFC = r'../data/cluster_output.shp'
outK = SSDO.CandidateField('KMEANS', 'LONG', groups + 1)
outMax = SSDO.CandidateField('MAXP', 'LONG', maxpGroups + 1)
outSKATER = SSDO.CandidateField('SKATER', 'LONG', skater.partitionOutput)
outFields = {'KMEANS': outK, 'MAXP': outMax, 'SKATER': outSKATER}
appendFields = fieldNames + ["NEW_NAME"]
ssdo.output2NewFC(outputFC, outFields, appendFields = appendFields)


# In[ ]: