While researches on the choice probability that a consumer chooses shopping sites are vast, little attention has been paid on the research to forecast the frequency of visitors at shopping sites. The first author has developed elsewhere the Markov shoparound model to forecast the number of visitors at shopping sites in the city center retail environment where the multipurpose, multistop trip chaining of consumers occurs as their shop-around behavior. The model was constructed according to the natural causal chain in which the entry frequency over shopping sites determines through consumers' shop-around the total number of visitors at each shopping site, which includes both of visitors who visit first there and who visit there after visiting other sites. However, to get the estimates of the unknown entry frequency is the most difficult task while large retailer usually count the number of visitors to their own facilities. With this in mind, this paper addresses the inverse problem to estimate the entry frequency from the observed number of visitors and shows that our proposed method based on the I-projection modeling under incomplete data is effective and accurate enough to estimate the entry frequency and consumers' shop-around pattern.