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

# # Creating and Using Panel Data Using ArcGIS Defined Location Cubes

# In[1]:


import os as OS
import arcpy as ARCPY
import SSDataObject as SSDO
import SSPanelObject as SSPO
import SSPanel as PANEL
ARCPY.overwriteOutput = True


# ## Example: Per Capita Incomes Relative to National Average in California (1969 - 2010)
# 
# - Create Defined Locations Cube from Repeating Shapes Feature Class
# - Run Mann-Kendall Trend Statistic
# - Save to NetCDF File

# In[2]:


inputFC = r'../data/CA_Counties_Panel.shp'
outputCube = r'../data/CA_Panel.nc'
fullFC = OS.path.abspath(inputFC)
outputCube = OS.path.abspath(outputCube)
fullPath, fcName = OS.path.split(fullFC)
ssdo = SSDO.SSDataObject(inputFC)
uniqueIDField = "MYID"
timeField = "YEAR"
analysisField = "PCR"
panelObj = SSPO.SSPanelObject(inputFC)
requireGeometry = panelObj.ssdo.shapeType.upper() == "POLYGON"
panelObj.obtainData(uniqueIDField, "YEAR", "1 Years", fields = [analysisField],
                    requireGeometry = requireGeometry)
panelCube = PANEL.SSPanel(outputCube, panelObj = panelObj)
varName = panelObj.fieldNames[0]
panelCube.mannKendall(varName)
panelCube.close()


# ## Open Panel Cube From NetCDF File for Analysis 

# In[3]:


panel = PANEL.SSPanel(outputCube)


# ## Number of Locations and Time Periods

# In[4]:


print("# locations = {0}, # time periods = {1}".format(panel.numLocations, panel.numTime))


# ## List Variables

# In[5]:


print(panel.obtainVariableList())


# ## View Mann-Kendall Trend Results in PANDAS Data Frame

# In[6]:


import pandas as PANDAS
locations = panel.locationLabel[0]
z = panel.obtainValues('PCR_TREND_ZSCORE')
pv = panel.obtainValues('PCR_TREND_PVALUE')
d = {'PCR_TREND_ZSCORE':z, 'PCR_TREND_PVALUE':pv}
df = PANDAS.DataFrame(d, index = locations)
print(df.head())


# ## Get 3D Analysis Variable 

# In[7]:


data = panel.obtainValues(analysisField)
print(data.shape)


# ## Use PySAL to Analyze LISA Markov Transitions

# In[8]:


import pysal as PYSAL
w = PYSAL.open(r"../data/queen.gal").read()
lm = PYSAL.LISA_Markov(data.T, w)
print(lm.classes)


# ## View Transistion Probabilities

# In[9]:


print(lm.p)


# In[10]:


panel.close()


# In[ ]: