When expanding a business, especially opening a new store in a new region, the biggest questions to be answered is whether the store can be expected to perform well and how much revenue it could potentially generate. Revenue prediction is very important in supporting site selection and decision making for ensuring a successful expansion, and there are several approaches that aim to deliver accurate forecasts.

Two machine learning algorithms that are commonly used to predict revenue are Random Forest and XGBoost. In this revenue prediction case study we compare the results from both these popular models with the results generated by Targomo’s Geo AI model. Geo AI is a location-based model, aiming for a better understanding and evaluation of the characteristics of the shop location.

The basis for this case study is the publicly available data for liquor sales data in the US-state Iowa. For the prediction, we selected the brand Hy-Vee and treated liquor sales of other brands as competitor data.

Who is Hy-Vee? Hy-Vee is an employee-owned chain of supermarkets, mostly located in the Midwestern and Southern US. Its headquarters is in West Des Moines, Iowa. Hy-Vee has over 285 locations in 8 states in the US, including among others Iowa, Illinois, Missouri. As all data of liquor sales in Iowa, Hy-Vee’s data was publicly available.

The three models Random Forest, XGBoost, and Targomo’s Geo AI were used for the revenue prediction. The Random Forest and XGBoost models are often applied in prediction as their advantage is their efficiency in dealing with high dimensional data. In addition, both random forest and XGBoost are decision-tree-based models and easy to work with – there’s no need to normalize data and Python/R packages are readily available to use. It can create a relative robust model even with outliers and missing values. The drawbacks, however, are also obvious: Potential over fitting issues and low interpretability.

Geo AI is a locational-based method. The foundations of the model were developed as part of a research project conducted by Targomo in collaboration with the Hasso-Plattner-Institut and funded by the Deutsches Zentrum für Luft- und Raumfahrt.

One major component of Geo AI is a gravitational model. It follows the idea of Newton’s law of gravitation applying the concept to market models. According to Newton’s law of gravitation, high masses (attractiveness in market models) and short distance lead to strong attraction. The attractiveness is measured by the store’s characteristics and environment. These factors could be characteristics like store size, parking space, product range, or location factors like complementary shops, competitors or other points of interest close by.

In our model, these factors are called attraction strength. The distance, in Geo AI, is calculated based on travel time with catchment areas defined by travel time and travel mode including car, bike, public transit, and walking. A travel-time-based catchment area provides a more accurate and realistic assessment of population than more naïve distance-based catchments.

To begin, Targomo’s data team first needed to compile data from various sources to calculate the general demand and attractiveness of stores:

Data cleaning is one of the first and most important steps in modelling as cleaned data is the foundation of a correct model and its output. The process of data cleaning involves detecting inaccurate, incomplete, and duplicated records, and then fixing or removing them.

Furthermore, stores with extreme high or low annual sales have also been carefully reviewed, noting that some were beverage and liquor retailers. Since these stores cater to different customer groups – namely other stores – these outliers were removed from the data set. Casinos, hotels, inns, and distilleries were also removed.

In countless iterations, the Geo AI engine is fed with data to continuously learn which data composition represents Hy-Vee’s store performance.

The training consists of two main components:

There are two relevant values measured during the learning process that indicate the prediction quality: The training error and the testing error.

When it comes to prediction results, the testing error can be considered to be more meaningful because it gives us an idea of how well the model will perform on new, unseen data – like the address of a potential new location for a shop opening. A low training error is a precondition for, but alone not a sufficient indicator of a successful project.

In our revenue prediction projects we usually go through two phases during the model training:

The table compares the model accuracy and detected success drivers of three liquor sales prediction models for Hy-Vee stores. The training error of Geo AI model is 16% and the testing error is 19.8%, while the testing errors of Random Forest model and XGBoost model are 38.1% and 30% respectively. Compared with the two commonly used models, Geo AI yields the highest model accuracy.

The identified attributes in the Random Forest and XGBoost models are rather limited and are restricted to the number of items, the number of visitors, and one or two socio-demographic attributes.

In contrast, the Geo AI model consists of a wide range of features, including among others store features (item count, shop type), surrounding environment (footfall at the location), and socio-demographic characteristics of potential customers. Item count, indicating the variety of liquors offered in store, shop type, and shopping footfall in the close neighbourhood are used to measure store attractivity in the gravitational model.

The scatter plot of the Geo AI output: The x-axis is the actual liquor sales, and the y-axis is the predicted liquor sales. The dots are nicely distributed along the diagonal line, which indicates a good prediction performance of Targomo’s Geo AI.

Income is a major driver of the liquor sales in Hy-Vee shops. 25.8% of the sales are from households with income less than 40,000$, and 14.5% are from households with incomes greater than 150,000$. Many studies have found that low-income individuals are at higher risk of engaging in heavy and hazardous drinking, and higher income is associated with a higher frequency of light drinking.

In addition to income, certain occupations have also a positive influence on liquor sales in Hy-Vee stores in Iowa. The results show that people working in construction, wholesale trade, and art industries as well as those who are self-employed are strong drivers. According to Statista, the top 5 high alcohol consumption (15 or more alcoholic drinks per week) occupations in the US 2016 are construction or mining, installation or repair, farming, fishing, or forestry, manufacturing or production, and business owner.

Among them, construction is one of the identified drivers in the model. The other identified occupations do also correlate strongly with the groups identified by our Geo AI, e.g. installation or repair, farming, fishing/forestry, manufacturing/production are often self-employed, as business owners are as well. So we can see that some drivers from Geo AI align with empirical results from research studies.

Integrated in the platform TargomoLOOP, the Geo AI model delivers immediate prediction result for any potential address.

Insight #1 Good estimate of attraction strength is key 

Attraction strength indicates how attractive our stores are and how attractive/competitive our competitors are. A good estimate of attraction strength can better estimate the market size and potential customers, and thus predict the revenue more accurately. In estimating the attraction strength, information about the store is key, for example, store size, sales area, store opening hours. In this case study, due to the limitation of public data, the information we have is not enough to capture the whole picture of the store attractiveness, such as the missing information about the store size. To circumvent this limitation, the building roof top data is used instead, but it is not possible to obtain this information for all stores.

Insight #2 An attractive shop location is defined by its surrounding business environment AND access to its target group

When selecting a new location to open a store, investigating the surrounding businesses is critical. What products/services do they offer? What is their vibe? Are they competitive or complementary? However, the geographic distribution of target demographics can be missed in the decision-making process. It does not matter how well suited a location is to a store if the target customer group are unable or unwilling to travel to it.

Using Random Forest and XGBoost models, it is hard to control what variables are used in the decision tree. In this case study, these two models mostly focused on the nearby environment while giving only minor consideration to the access to target demographics. The Geo AI model both put a greater emphasis on demographics, and took into account a wider array of demographic and environmental factors. We have abundant data to measure various aspects of demographic structure to support the model and future decision making.

Insight #3 Expert knowledge could be a double-edged sword in key driver analysis. 

When selecting variables in the driver analysis, expert knowledge can provide insights on what key drivers can be expected based on experience or existing literature. At the same time, relying too much on expert knowledge and excluding data at the beginning that we thought is unrelated can fall into a bias trap.

Instead, by letting the model and data speak, we can gain new or unseen insights. We may find some variable we thought could be important have little to no significance, and vice versa. What we should then do is to investigate it, figure out why and explain it. Our assumptions should not interfere with the model.

The feature selection mechanism in Geo AI ensures the model suffers less from overfitting even with large number of features as the input.

This case study tests and compares the performance of three models, Random Forest, XGBoost, and Geo AI, as revenue prediction methods, using liquor sales data from Iowa, USA. Geo AI has the best performance among all 3 models in predicting liquor sales for Hy-Vee, having the least overfitting, the lowest testing error, and the important features in the model are more comprehensive.

Compared with Random Forest and XGBoost models, Geo AI takes greater consideration of shop feature factor variables, shop surrounding environment, local demographic structure, and mobility data making the model more reliable. In addition, Targomo’s Geo AI, a locational-based model, offers better understanding and evaluation of the characteristics of the shop location. It is possible to track the model and retrieve the prediction of each store to identify key revenue drivers. Due to its rate of overfitting and testing error, Geo AI could better predict the revenue of a new location and analyse revenue drivers.

Liquor sales prediction is one example of a Geo AI model. Geo AI can be applied in various industries, including among others retail, delivery services, food industry, and health care. Are you interested to learn more about Targomo’s Geo AI forecasting for retailers and restaurant brands, and see dynamic analytics of the Hy-Vee Geo AI model? Book a demo here

Authors: Yue Luo (Spatial Data Scientist), Gideon Cohen (Software Engineer), Luisa Sieveking (Marketing & Communications)

The Geo AI model was developed by Targomo’s Data Science Team that is dedicated to evaluate which location factors influence the business success of brands and companies, and to what extend they contribute to revenue.

 “Find out how we used Isaac Newton’s Theory of Gravitation to answer a complex location based question: Why are so many Berlin parents struggling to find a Kindergarten place for their children? We found that Kindergartens throughout the city are unevenly distributed.  

Since 2018, every child in the German capital is entitled to free child care from their first birthday onward. Still, this new law has not solved the problem that has become very common among young parents. The threat of “Kita-Notstand” (“Kindergarten Emergency”) has recently made headlines. It threatens the attractiveness of Berlin for young parents and newcomers. And here at Targomo we have also felt this problem. After all, we are a young company and some of our co-workers just had children and some are struggling to find a place for their child. 

We were wondering: What is the story behind the Kita-Notstand? Is there maybe something we are missing here? As a first step we decided to look at the demand and supply of Kita places and it looks like there are enough places for children in the German capital. In total 139.856 kids are enrolled in Berlin’s Kitas today and the whole city offers 152.559 Kita places (destatis.de, kita-suche.de). 

So we asked our data magician Jacopo to further examine the problem from a location intelligence perspective. We assumed that the locations of Berlin’s Kitas were suboptimal. To test this idea he used a concept based on Newton’s law of gravitation particularly helpful in solving this puzzle.

Newton’s law of gravity says: “every mass attracts every other mass in the universe, and the gravitational force between two bodies is proportional to the product of their masses, and inversely proportional to the square of the distance between them.” In short: The gravitational force between two bodies gets smaller, the further apart they are. And larger bodies have a stronger gravitational force than smaller bodies.

Our model treats Kita locations as bodies that attract people. Larger Kitas attract more children. The longer people must travel to a Kita, the less they are attracted to it. With our gravitational model we can predict the number of kids applying to each Kita and thus their degree of overcapacity.

Very early in the process we decided to follow two strands of analysis. First, we wanted to see what would happen if parents chose the closest Kita to their homes? Second, what if all kids eligible for a Kita spot applied for it? How would this change the current picture?

But in order to get the full picture we needed to focus on two crucial categories: Age groups and location. A Kita in your neighborhood is not helpful if it only serves kids older than 3 years, but your child is 2-year old. A Kita with a free space for your child that is situated at the other end of the city is similarly unrealistic.

To understand the two factors better, let’s take a look at the following maps. The visualization of our gravitational analysis reveals the problems in the distribution of Kita places across the city. Black dots represent Kitas that are over capacity, meaning there are more Kids in its surroundings than the Kita can service. The size of the dots signifies the degree of overcapacity; how many children a Kita would have to accommodate to serve the demands of the parents living close by. Red dots equal 75% occupancy. Comparing these maps shows that the picture varies drastically between age groups.

Kids younger than 3, especially in the South-West of the city, face a dire situation. For kids older than 3, there are enough Kita spots but there is a slight mismatch. Kitas in the East are on average only 75% occupied while Kitas in the West and at the outskirts of the city cannot serve all the children in their proximity. We calculate that 52 % of Kitas serving children between 1 and 3 years old are too small for the demand in their direct proximity. Or to put it differently, 48% of Kitas do not have enough children living in their proximity, that means that parents have to travel to these Kitas to find a place for their children. This situation is less dire for older children but still 30% of Kitas are over capacity.

In Berlin, every child over the age of 1 is entitled to a place in a Kita. The question we were asking is: what if every family in Berlin claimed their right to a Kita place? The result is dramatic.  According to our census-based analysis, today about 35.000 kids in this age group who have a right to a Kita, don’t have access to a place.

The picture that emerges makes the problem crystal clear:

Children are the future. Providing equitable access to Kita benefits the entire community.

The reality on the ground probably resembles a state that’s somewhere in the middle of the two cases we considered. Giving easier access to Kitas in close vicinity to parents homes would incentivize more parents sending their children to a Kita. On the other hand, we know that not every parent will send their child to the Kita for a variety of reasons. But nobody wants a world where parents are forced to quit their jobs because the only available Kita is too far away from their home.

We understand that there are many more layers of complexity to any public policy decision that make running a city that much more complex. But we believe that visualizing the scope of the problem is incredibly powerful as a starting point for policy. With our Location Intelligence solutions, we give decision makers a platform to make decisions that are informed by data and state-of the art technology.

The study was presented by Jacopo Solari at the Urban Mobility Sympoisum – Karten, Daten, Geovisualisierung on 11. Oktober 2019 at CityLAB Berlin, organized by Technologiestiftung Berlin, supported by Senatskanzlei Berlin.

As we support the public sector, we participated in the Smart Country Convention Berlin last week. This forum for the exchange of strategic advice provided the opportunity to learn more from experts about the current state of affairs. While most cities are facing unique situations on a granular level, many places struggle with similar challenges. This year’s partnering country – and role model – was Lithuania. For them, becoming smart was the most pragmatic choice: “We’re not big, we’re not rich, so we have to be smarter”, their Foreign Minister Linas Antanas Linkevičius explained.

The impressive Baltic state has been so successful in fact, that its capital city Vilnius is overwhelmed by the incoming UK tech companies looking for new places to run their business. The city’s recipe for success consists of both top-down and bottom-up approaches. By creating the optimal legal frameworks and infrastructure conditions, the government paves the way for young companies to thrive with new business models.

Some of these companies are working towards new transport solutions, with good reason: “Regardless of size, all cities have mobility issues”, the mayor of Vilnius, Remigijus Šimašius said on stage. Once citizens establish routines, adjustments are difficult because of inherent resistance to changing habits (individual car ownership for example).

The mayor reminded us that the main beneficiaries of mobility are neither the municipalities nor the service providers, but the people. Mobility should not be an end in itself. Instead, mobility solutions should focus on the benefits they bring to their users.

The starting point of innovations should be micro-decisions of citizens as opposed to some overarching goal. But how to tackle all these challenges? One way, Vilnius decided, was a radical open data policy. Vilnius claims to be one of the first in Europe to create a public city data platform that provides open-source data for all businesses and citizens – “without excuses”.  As a consequence, since September 2017 it’s the home of Trafi, one of the most advanced mobility apps in the world. Trafi bundles all public mobility solutions, making it easy to navigate the city without a private car. This is not a one-way success story tough. Trafi provides the city administration with so much new data that it’s fundamentally changing traffic planning.

Other cities are also impressed: Recently, Trafi and the BVG (Berlin’s transit agency) launched Jelbi, to bring the same mobility experience to the citizens of the German capital.

But if igniting the spark is so easy, why are many city governments struggling to follow that open data approach? In our daily work with open data sources from all over the world, we encounter a wide range of problems. Two of the most significant ones: Coverage and quality.

For many of our customers, precise travel time analysis with public transport is key for answering their mobility questions. Integrating public transport data for Germany alone means using data sets from more than 80 federal transport authorities, many of which have to be contacted personally and may be reluctant to share their data. Consequently, it’s almost impossible to create a complete picture of Germany’s public transport landscape at this moment.

On the other hand, good intentions on their own aren’t enough, the correct execution is equally important. Large-scale data sets need homogeneous structures. Lina Bruns, a research assistant at Fraunhofer FOKUS, demonstrated at the Smart Country Convention how minor syntactical inconsistencies in data descriptions – such as putting a phone number as either 040/9595 or 040-9595 – can cause considerable additional effort, while incomplete data sets can quickly render all available information unusable for the majority of analytical purposes.

Bruns argued that agencies lack strategy, processes, tools, or resources to make sure their data solutions fulfill the necessary quality requirements. To help them, FOKUS released new guidelines this week for the improvement of data and meta-data quality, focusing especially on the necessary setup and structure of data.

While this is certainly a great help for those who already have an open data strategy, we hope that events like the Smart Country Convention convince more government agencies that they need one in the first place.