Need some help on how to read and interpret the results

[SaiAditya2595]

Clarification

I am new to geolift library and I have followed the exact same steps which are mentioned in the documentation page. Mine is a weekly transactional data in 17 locations. Initially I want to include "Arizona" as test market in my case. Got the output as below. Selected Arizona n Florida as test regions based on rank 2. Ran Geolift for the above selected test markets and got negative lift. If I select the two above mentioned states as test campaign markets, overall txns drop to -103359? How can I interpret this result as well as model weights such as * california: 0.6125 and * louisiana: 0.3875? Any help is greatly appreciated. Thanks a lot

MarketSelections <- GeoLiftMarketSelection(data = GeoTestData_PreTest, treatment_periods = c(2,4), N = c(2,3), Y_id = "Y", location_id = "location", time_id = "time", effect_size = seq(0, 0.25, 0.05), lookback_window = 5, include_markets = c("arizona"), #exclude_markets = c("California"), #holdout = c(0.5, 1), #cpic = 7.50, budget = 200000, #alpha = 0.1, Correlations = TRUE, #fixed_effects = TRUE, #side_of_test = "one_sided" ) ID location duration EffectSize Power AvgScaledL2Imbalance Investment 1 1 arizona, california, florida 2 0.25 1 0.6377641 181290.0 2 2 arizona, florida 2 0.25 1 0.4156887 104589.1 AvgATT Average_MDE ProportionTotal_Y abs_lift_in_zero Holdout rank correlation 1 23385.00 0.1830134 0.4689118 0.067 0.5310882 1 0.9285809 2 16023.89 0.1397430 0.2802650 0.110 0.7197350 2 0.9006815

GeoTest <- GeoLift(Y_id = "Y",

•                data = GeoTestData_Test,
  

•                locations = c("arizona", "florida"),
  

•                 #c("111-sacramento", "113-san francisco"),
  

•                treatment_start_time = 48,
  

•                treatment_end_time = 52)
  

GeoTest

GeoLift Output

Test results for 5 treatment periods, from time-stamp 48 to 52 for test markets: 1 ARIZONA 2 FLORIDA ##################################

Test Statistics

##################################

Percent Lift: -9%

Incremental Y: -103359

Average Estimated Treatment Effect (ATT): -10335.897

The results are significant at a 95% level. (TWO-SIDED LIFT TEST)

There is a 1.8% chance of observing an effect this large or larger assuming treatment effect is zero.

summary(GeoTest,table=TRUE)

GeoLift Results Summary

##################################

Test Statistics

##################################

• Average ATT: -10335.897
• Percent Lift: -9%
• Incremental Y: -103359
• P-value: 0.02

##################################

Balance Statistics

##################################

• L2 Imbalance: 47508.654
• Scaled L2 Imbalance: 0.4051
• Percent improvement from naive model: 59.49%
• Average Estimated Bias: NA

##################################

Model Weights

##################################

• Prognostic Function: NONE

• Model Weights:

• california: 0.6125

• louisiana: 0.3875