GEOG 371 Lecture Notes - Lecture 15: Autocorrelation, Statistical Significance, Scatter Plot
5 May 2018
School
Department
Course
Professor
GEOG 371 Final Exam Review
Week 15
4/23 – Spatial Autocorrelation for Interval Variables
Moras I
• Correlation coefficient
• Interval-scale variable
• Type and statistical significance of spatial autocorrelation
• (Part of) equation:
o Wij = spatial weight location for location i and location j
▪ = 1 if i and j are close, 0 otherwise
o I > 0 → positive SA
o I = 0 → o SA rado
o I < 0 → egatie SA
• Expected value of I:
o
o N = # of areas
Z- Test
• Ho: Moras I = EI
• Ha: Moras I ≠ EI
•
o This approach is not recommended
• Better approach to test statistical significance
o Monte Carlo Randomization
▪ 1 – create large # of random values of I > actual I
▪ 2 – for eah rado patter, alulate Moras I alue
▪ 3 – reate histogra of Moras I alues for rado patters
▪ 4 – loate Atual Moras I o histogra
• P-value = proportion of random values of I > actual I
• If p < 0.05, reject null, there is SA
Correlation Coefficient Plot
• Wx = avg value of X in neighboring areas
4/25 – Local Spatial Autocorrelation (LISA)
LISA
• Loal ersio of Moras I
• Used to identify clusters – hot spots, cold spots
• High, positive value indicates clustering
find more resources at oneclass.com
find more resources at oneclass.com
