test_fake_lift {MarketMatching}

Given a test market, analyze the impact of fake interventions (prospective power analysis)

Description

test_fake_lift Given a matched_market object from the best_matches function, this function analyzes the causal impact of fake interventions using the CausalImpact package. The function returns an object of type "market_inference" which contains the estimated impact of the intervention (absolute and relative).

Usage

test_fake_lift(matched_markets=NULL,
          test_market=NULL,
          end_fake_post_period=NULL,
          alpha=0.05,
          prior_level_sd=0.01,
          control_matches=NULL, 
          nseasons=NULL, 
          max_fake_lift=NULL, 
          steps=10,
          lift_pattern_type="constant")

Arguments

matched_markets

A matched_market object created by the market_matching function This parameter will overwrite the values specified in prior_level_sd and nseasons. ONLY use this if you're using intricate bsts settings For most use-cases, using the prior_level_sd and nseasons parameters should be sufficient

test_market

The name of the test market (character)

end_fake_post_period

The end date of the post period. Must be a character of format "YYYY-MM-DD" – e.g., "2015-11-01"

alpha

Desired tail-area probability for posterior intervals. For example, 0.05 yields 0.95 intervals

prior_level_sd

Prior SD for the local level term (Gaussian random walk). Default is 0.01. The bigger this number is, the more wiggliness is allowed for the local level term. Note that more wiggly local level terms also translate into larger posterior intervals This parameter will be overwritten if you're using the bsts_modelargs parameter

control_matches

Number of matching control markets to use in the analysis (default is 5)

nseasons

Seasonality for the bsts model – e.g., 52 for weekly seasonality

max_fake_lift

The maximum absolute fake lift – e.g., 0.1 means that the max lift evaluated is 10 percent and the min lift is -10 percent Note that randomization is injected into the lift, which means that the max lift will not be exactly as specified

steps

The number of steps used to calculate the power curve (default is 10)

lift_pattern_type

Lift pattern. Default is constant. The other choice is a random lift..

Value

Returns an object of type matched_market_power. The object has the following elements:

ResultsData

The results stored in a data.frame

ResultsGraph

The results stored in a ggplot graph

LiftPattern

The random pattern applied to the lift

FitCharts

The underlying actual versus fitted charts for each fake lift

FitData

The underlying actual versus fitted data for each fake lift

Examples

## Not run: 
library(MarketMatching)
##-----------------------------------------------------------------------
## Create a pseudo power curve for various levels of lift
## Since this is weather data it is a not a very meaningful example. 
## This is merely to demonstrate the function.
##-----------------------------------------------------------------------
data(weather, package="MarketMatching")
mm <- best_matches(data=weather, 
                   id="Area",
                   markets_to_be_matched=c("CPH", "SFO"),
                   date_variable="Date",
                   matching_variable="Mean_TemperatureF",
                   warping_limit=1, # warping limit=1
                   dtw_emphasis=0, # rely only on dtw for pre-screening
                   matches=5, # request 5 matches
                   start_match_period="2014-01-01",
                   end_match_period="2014-10-01")
library(CausalImpact)
results <- test_fake_lift(matched_markets=mm,
                     test_market="CPH",
                     lift_pattern_type="constant",
                     control_matches=5, # use all 5 matches for inference
                     end_fake_post_period="2015-12-15",
                     prior_level_sd=0.002, 
                     max_fake_lift=0.1)

## End(Not run)