Client Background

Motivation & Use Case

Bicycle count data is important for a variety of planning purposes, including measuring changes over time, prioritizing projects, planning and designing for future infrastructures, providing modelling inputs for multi-modal models, assessing overall footfall, signal adjustmetns and property research among other things (Nordback et al 2018). The growing interest in bicycling and cycling facilities necessitates a need to understand its usage and how people utilize the mode and available facilities for getting to important places in their daily life. This need is not unnoticed by the DDOT, who has been carrying out their own data collection program for the past few years. However, as we further reveal in subsequent sections, there are limits to the ability of collected data to generalize to the entire city and its network.

Essentially addressing a problem of incomplete information, our analysis will estimates and provides summaries of micro-mobility ridership across DC’s networks. By revealing the distribution and gaps in available data, our study highlights opportunities to enhance monitoring strategies, understand how patterns vary across the city with various demographic and infrastructure features, and guide targeted infrastructure improvements. This is facilitated by a dashboard which provides various functions for exploring the phenomena to aid decision making processes.

The Role of Bicycle Count Data

Bicycle count data serve various planning and evaluation purposes, including measuring trends, prioritizing infrastructure investments, calibrating travel models, conducting before-and-after studies, and assessing safety and equity outcomes (Nordback et al., 2018). However, stationary counters can only reflect activity at specific points, leaving large portions of the network unmeasured. To address this, researchers have explored methods to extrapolate short-duration counts to average annual daily bicycle traffic (AADBT) using grouping techniques such as cluster analysis, spatial indexing, and travel pattern similarities (Team 2025).

Modeling Approaches

Connectivity and Network Measures

Complementing count-based models, Miah et al. (2023) explored network connectivity using graph theory and Level of Traffic Stress (LTS). Their study assessed how link-level characteristics—such as speed, lane width, and motor vehicle volume—affect ridership by user type (e.g., children vs. confident adults). The results highlighted significant connectivity disparities, with large portions of networks inaccessible to low-stress riders, thereby limiting cycling uptake in equity-seeking communities (Miah, Mattingly, and Hyun 2023).

Our Target Variable: Weekday Hourly Bike Counts

manual <- manual %>% drop_na(x, y)
manual <- st_as_sf(manual, coords = c("x", "y"), crs = 4326) 
manual <- st_transform(manual, st_crs(basemap))
ggplot() +
  geom_sf(data = basemap, fill = "grey90", color = "white") +
  geom_sf(data = bike_lane, color = "grey30", size = 0.5, alpha = 0.7) +
  geom_sf(data = manual, color = "#E3D7F5", size = 3, shape = 21, fill = "#562DAB") +
  labs(title = "Manual Counters Locations") +
  theme_minimal() +  
  theme(
    legend.position = "right",   
    panel.grid.major = element_blank(), 
    panel.grid.minor = element_blank(),  
    axis.text = element_blank(), 
    axis.ticks = element_blank())

auto <- auto %>% drop_na(Latitude, Longitude)
auto <- st_as_sf(auto, 
                  coords = c("Longitude", "Latitude"),  # note the order: X = Longitude, Y = Latitude
                  crs = 4326)
auto <- st_transform(auto, st_crs(basemap))

auto_cut <- auto %>%
  filter(st_coordinates(.)[, 1] > -77.12 & st_coordinates(.)[, 1] < -76.90,
         st_coordinates(.)[, 2] > 38.79 & st_coordinates(.)[, 2] < 39.00)
ggplot() +
  geom_sf(data = basemap, fill = "grey90", color = "white") +
  geom_sf(data = bike_lane, color = "grey30", size = 0.5, alpha = 0.7) +
  geom_sf(data = auto_cut, color = "#E3D7F5", size = 3, shape = 21, fill = "#9C7BD8") +
  labs(title = "Automatic Counters Locations") +
  theme_minimal() +  
  theme(
    legend.position = "right",    
    panel.grid.major = element_blank(),   
    panel.grid.minor = element_blank(),  
    axis.text = element_blank(),   
    axis.ticks = element_blank())

Our Approach: Estimate Spatial Patterns with Manual Counts & Temporal Patterns from Automated Counts

Typical approaches for obtaining estimates of bike counts in areas without counters can be classified into three main types, 1) Direct Extrapolation , 2) Travel Demand Modelling and 3) Direct Demand Models.

Direct extrapolation are based on the principles of spatial autocorrelation, that areas near each other have similar patterns of counts. This often entails grouping sites based on similar characteristics and using factors to extrapolate short duration counts to annual estimates. This may be done visually, or through methods of cluster analysis or spatial variables.

Travel demand modelling describes patterns as decisions to travel for activities using specific modes of transport, and often suffer from their large aggregation scale and difficulty in validating with external data.

As such, given the availability of various external data sources, direct demand models, which regresses (or predicts) observed counts based on known characteristics of the surrounding built environment, are popular methods for estimating counts. The accuracy of such models depends on the types of variables selected as predictors, as well as their suitability for extrapolating to the entire network - meaning that including strongly predictive characteristics not present within areas without records should be approached with caution.

Crucially, majority of reviewed approaches are often concerned with obtaining annually aggregated estimates of bicycle volume, in other words, where there tends to be most or least cyclists on the network. Our use-case and motivation, however, requires that we also understand when and how bike counts vary.

Given the consistency of hourly patterns across locations, the wealth of temporal data from automated counters allows us to estimate how patterns vary across the day, barring a few limitations and caveats which are explained further below.

As such, we propose a modelling workflow which can maximize the degree of spatial and temporal information we can obtain from the 2 data sources.

Spatial Model

Variables chosen for the spatial model included….

Temporal Model

Given that our approach means that we can not utilize accurate spatial grouping variables (e.g streets) to provide information regarding the For the temporal model, we considered the usage of various spatial variables for providing information regarding the spatial differences in the magnitude of bicycle counts. This included predicted values based of the spatial model, a kernel density estimate of manual count data and engineered count feature from capital bikeshare data.

Capital Shared Bike Data

Our analysis reveals a characteristic “twin-peak” trend in cycling activity, closely mirroring the morning and evening commuter surges seen in other transport modes. This pattern highlights how bike usage aligns with standard workday rhythms, suggesting that cycling also plays a meaningful role during peak travel times.

By incorporating these repeated daily peaks into our modeling approach, we can better predict where and when people are most likely to ride. Understanding how traffic volume shifts at specific locations allows us to fine-tune infrastructure, plan more effectively for congestion, and ultimately deliver a safer, more efficient cycling experience throughout Washington, D.C.

process_auto_counts <- function(auto_counts) {
    
    auto_counts$interval_60 <- as.POSIXct(auto_counts$interval_60, origin="1970-01-01", tz="UTC")
    auto_counts$hour <- hour(auto_counts$interval_60)
    auto_counts$day <- as.Date(auto_counts$interval_60)
    auto_counts$weekday <- wday(auto_counts$interval_60, week_start = 1)
    auto_counts$is_weekend <- auto_counts$weekday %in% c(6, 7)
    return(auto_counts)
}

auto_counts <- process_auto_counts(auto_counts)
process_cabi_data <- function(cabi_data) {
    cabi_data$hour <- ifelse(is.na(cabi_data$hour), -1, as.integer(cabi_data$hour))
    cabi_data$dotw <- ifelse(is.na(cabi_data$dotw), -1, as.integer(cabi_data$dotw))
    cabi_data$is_weekend <- cabi_data$dotw %in% c(6, 7)
    return(cabi_data)
}
cabi_data <- process_cabi_data(cabi_sum)

plot_hourly_patterns <- function(auto_counts, cabi_data) {
    hourly_auto <- auto_counts %>% group_by(hour) %>% summarize(Flow_Count = mean(Flow_Count, na.rm=TRUE))
    hourly_cabi <- cabi_data %>% group_by(hour) %>% summarize(cabi_count = mean(cabi_count, na.rm=TRUE))
    
    p1 <- ggplot(hourly_auto, aes(x = hour, y = Flow_Count, fill = "Auto Counts")) + 
        geom_bar(stat="identity", color = "#C4AEE8") + 
        scale_fill_manual(values = c("Auto Counts" = "#562DAB")) + 
        ggtitle("Average Hourly Auto Counts") + 
        xlab("Hour of Day") + 
        ylab("Average Count") + 
        theme_minimal()

    p2 <- ggplot(hourly_cabi, aes(x = hour, y = cabi_count, fill = "CABI Usage")) + 
        geom_bar(stat="identity", color = "#C4AEE8") + 
        scale_fill_manual(values = c("CABI Usage" = "#6A5ACD")) + 
        ggtitle("Average Hourly CABI Usage") + 
        xlab("Hour of Day") + 
        ylab("Average Count") + 
        theme_minimal()

weekday_auto <- auto_counts %>% group_by(hour, is_weekend) %>% summarize(Flow_Count = mean(Flow_Count, na.rm=TRUE))

weekday_cabi <- cabi_data %>% group_by(hour, is_weekend) %>% summarize(cabi_count = mean(cabi_count, na.rm=TRUE))

    p3 <- ggplot(weekday_auto, aes(x = hour, y = Flow_Count, color = factor(is_weekend))) + 
        geom_line(size = 1.2) + 
        scale_color_manual(values = c("FALSE" = "#C4AEE8", "TRUE" = "#6A5ACD")) + 
        ggtitle("Weekday vs Weekend - Auto Counts") + 
        xlab("Hour of Day") + 
        ylab("Average Count") + 
        theme_minimal()

    p4 <- ggplot(weekday_cabi, aes(x = hour, y = cabi_count, color = factor(is_weekend))) +
        geom_line(size = 1.2) + 
        scale_color_manual(values = c("FALSE" = "#C4AEE8", "TRUE" = "#6A5ACD")) + 
        ggtitle("Weekday vs Weekend - CABI Usage") + 
        xlab("Hour of Day") + 
        ylab("Average Count") + 
        theme_minimal()
    return(list(p1, p2, p3, p4))
}

analyze_peak_hours <- function(auto_counts, cabi_data) {
    morning_peak <- 7:9
    evening_peak <- 16:18

    auto_peaks <- data.frame(
        morning = mean(auto_counts$Flow_Count[auto_counts$hour %in% morning_peak], na.rm=TRUE),
        evening = mean(auto_counts$Flow_Count[auto_counts$hour %in% evening_peak], na.rm=TRUE),
        off_peak = mean(auto_counts$Flow_Count[!(auto_counts$hour %in% c(morning_peak, evening_peak))], na.rm=TRUE))

    cabi_peaks <- data.frame(
        morning = mean(cabi_data$cabi_count[cabi_data$hour %in% morning_peak], na.rm=TRUE),
        evening = mean(cabi_data$cabi_count[cabi_data$hour %in% evening_peak], na.rm=TRUE),
        off_peak = mean(cabi_data$cabi_count[!(cabi_data$hour %in% c(morning_peak, evening_peak))], na.rm=TRUE)
    )
    return(list(Auto_Counts = auto_peaks, CABI_Usage = cabi_peaks))
}

plots <- plot_hourly_patterns(auto_counts, cabi_data)
peak_stats <- analyze_peak_hours(auto_counts, cabi_data)

print(plots[[2]])

print(plots[[4]])

Land Use

To understand how land use influences bike counts, we applied a 300-meter buffer around each bike lane segment. This allows us to capture the mix of land uses within close proximity to each bike route, rather than just relying on a single adjacent land use type.

The highest bike counts are observed in residential, commercial, and institutional areas, indicating that these environments support frequent cycling—likely due to a mix of commuting, errands, and recreational trips.

seg_nongem <- st_drop_geometry(segments)
landuse <- seg_nongem %>%
  select(Commercial, Industrial, Institutional, Mixed, Recreational, Religious, Residential, Vacant, Other) %>%
  summarise_all(sum) %>%
  pivot_longer(cols = everything(), names_to = "Land_Use_Type", values_to = "Count")

ggplot(landuse, aes(x = reorder(Land_Use_Type, Count), y = Count, fill = Land_Use_Type)) +   
  geom_col(width = 0.7) +  # Adjust bar width (default is 0.9, reduce for thinner bars)
  coord_flip() +   
scale_fill_manual(values = c(
  "Vacant"        = "#562DAB",  # Very dark purple
  "Residential"   = "#6A3BB8",
  "Institutional" = "#7A50C7",
  "Commercial"    = "#8E68CC",
  "Religious"     = "#9C7BD8",
  "Mixed"         = "#A989E0",
  "Other"         = "#C4AEE8",
  "Industrial"    = "#D6C4F0",
  "Recreational"  = "#E3D7F5"   # Lightest
), guide = "none") +  
  labs(title = "Distribution of Bike Lanes Across Adjacent Land Use Types",
       x = "Land Use Type", y = "Count") +   
  theme_minimal() +
  theme(
    legend.position = "right",   # Keep legend on the right
    panel.grid.major = element_blank())

Census Data

We examined the spatial distribution of key census indicators across our study area to better understand the local socio-demographic context. Our maps reveal marked differences in variables such as median income, the rate of public transit commuting, and the proportion of residents who use bicycles for commuting. Areas with higher median incomes may have distinct mobility patterns compared to regions with lower incomes, and these economic disparities can influence the demand for cycling infrastructure. By overlaying these data on our geographic base map, we can begin to see how the underlying socio-economic factors correlate with the spatial distribution of transportation modes.

In addition, the visualizations of transit and bicycle commuting rates serve as essential proxies for local travel behavior. Regions with higher bicycle commuting rates, for example, might already exhibit a strong cycling culture that can support further network investments, while areas with prominent public transit usage suggest the potential for integrated transport solutions that merge cycling with existing transit services. These insights are critical for identifying data gaps in our current bike counter coverage and for forecasting future demand, ultimately guiding more targeted and effective infrastructure planning in Washington, D.C.

ggplot() +
  geom_sf(data = basemap, fill = "grey90", color = "white") +
  geom_sf(
    data = segments,
    aes(color = median_incomeE),
    alpha = 0.8) +
    scale_color_gradientn(
    colors = rev(pp),
    name = "Income",
    na.value = "grey80") +
  labs(title = "Median Income") +
  theme_minimal()

ggplot() +
  geom_sf(data = basemap, fill = "grey90", color = "white") +
  geom_sf(
    data = segments,
    aes(color = commute_bicycleE),
    alpha = 0.8) +
    scale_color_gradientn(
    colors = rev(pp),
    name = "commute_bicycle") +
  labs(title = "Bicycle Commuters") +
  theme_minimal()

ggplot() +
  geom_sf(data = basemap, fill = "grey90", color = "white") +
  geom_sf(
    data = segments,
    aes(color = pct_white),
    alpha = 0.8) +
    scale_color_gradientn(
    colors = rev(pp),
    name = "pct_white") +
  labs(title = "Percentage of White") +
  theme_minimal()

ggplot() +
  geom_sf(data = basemap, fill = "grey90", color = "white") +
  geom_sf(
    data = segments,
    aes(color = edu_bachelors_plusE),
    alpha = 0.8) +
    scale_color_gradientn(
    colors = rev(pp),
    name = " edu_bachelor_plus") +
  labs(title = "Education Level") +
  theme_minimal()

Network Characteristics

We also examine the network characteristics as our additional resources to help our model. Betweenness Centrality explains how often a node falls on the shortest path between other nodes.

ggplot() +
  geom_sf(data = basemap, fill = "grey97", color = "white") +
  geom_sf(
    data = segments,
    aes(color = betweenness_max),
    alpha = 0.8) +
    scale_color_gradientn(
    colors = rev(pp),
    name = "Betweenness") +
  labs(title = "Bike Lane Network: Binned Betweenness Centrality") +
  theme_minimal()

Exploratory Data Analysis with Manual Counts

In this section, we explore how various segment-level characteristics—such as roadway design, land use, and traffic conditions—relate to observed manual bicycle counts in Washington, DC. This exploratory analysis aims to identify potential predictors of cyclist volumes and inform our modeling efforts to estimate unobserved counts citywide.

segments_manual <- segments %>%
  semi_join(seg_manual, by = "seg_id") %>%     
  left_join(seg_manual, by = "seg_id")

Overview of Average Current Manual Counts

ggplot() +
  geom_sf(data = basemap, fill = "grey97", color = "white") +
  geom_sf(
    data = segments_manual,
    aes(color = avg_count),
    alpha = 0.8) +
    scale_color_gradientn(
    colors = rev(pp),
    name = "Average Bike Counts") +
  labs(title = "Manual Bike Counts") +
  theme_minimal()

ggplot(segments_manual, aes(x = avg_count)) +
  geom_histogram(bins = 30, fill = "#7A50C7", color = "white") +
  theme_minimal() +
  labs(title = "Distribution of Average Manual Counts",
       x = "Average Count", y = "Number of Segments")