Time Series Clustering
Clustering is an
important part of time series analysis that allows us to organize time
series into groups by combining “tsfeatures” (summary matricies) with
unsupervised techniques such as K-Means Clustering. In this short
tutorial, we will cover the tk_tsfeatures() functions that
computes a time series feature matrix of summarized information on one
or more time series.
Libraries
To get started, load the following libraries.
Data
This tutorial will use the walmart_sales_weekly
dataset:
- Weekly
- Sales spikes at various events
## # A tibble: 1,001 × 17
## id Store Dept Date Weekly_Sales IsHoliday Type Size Temperature
## <fct> <dbl> <dbl> <date> <dbl> <lgl> <chr> <dbl> <dbl>
## 1 1_1 1 1 2010-02-05 24924. FALSE A 151315 42.3
## 2 1_1 1 1 2010-02-12 46039. TRUE A 151315 38.5
## 3 1_1 1 1 2010-02-19 41596. FALSE A 151315 39.9
## 4 1_1 1 1 2010-02-26 19404. FALSE A 151315 46.6
## 5 1_1 1 1 2010-03-05 21828. FALSE A 151315 46.5
## 6 1_1 1 1 2010-03-12 21043. FALSE A 151315 57.8
## 7 1_1 1 1 2010-03-19 22137. FALSE A 151315 54.6
## 8 1_1 1 1 2010-03-26 26229. FALSE A 151315 51.4
## 9 1_1 1 1 2010-04-02 57258. FALSE A 151315 62.3
## 10 1_1 1 1 2010-04-09 42961. FALSE A 151315 65.9
## # ℹ 991 more rows
## # ℹ 8 more variables: Fuel_Price <dbl>, MarkDown1 <dbl>, MarkDown2 <dbl>,
## # MarkDown3 <dbl>, MarkDown4 <dbl>, MarkDown5 <dbl>, CPI <dbl>,
## # Unemployment <dbl>TS Features
Using the tk_tsfeatures() function, we can quickly get
the “tsfeatures” for each of the time series. A few important
points:
The
featuresparameter come from thetsfeaturesR package. Use one of the function names fromtsfeaturesR package e.g.(“lumpiness”, “stl_features”).We can supply any function that returns an aggregation (e.g. “mean” will apply the
base::mean()function).You can supply custom functions by creating a function and providing it (e.g.
my_mean()defined below)
# Custom Function
my_mean <- function(x, na.rm=TRUE) {
mean(x, na.rm = na.rm)
}
tsfeature_tbl <- walmart_sales_weekly %>%
group_by(id) %>%
tk_tsfeatures(
.date_var = Date,
.value = Weekly_Sales,
.period = 52,
.features = c("frequency", "stl_features", "entropy", "acf_features", "my_mean"),
.scale = TRUE,
.prefix = "ts_"
) %>%
ungroup()
tsfeature_tbl## # A tibble: 7 × 22
## id ts_frequency ts_nperiods ts_seasonal_period ts_trend ts_spike
## <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 1_1 52 1 52 0.000670 0.0000280
## 2 1_3 52 1 52 0.0614 0.00000987
## 3 1_8 52 1 52 0.756 0.00000195
## 4 1_13 52 1 52 0.354 0.00000475
## 5 1_38 52 1 52 0.425 0.0000179
## 6 1_93 52 1 52 0.791 0.000000754
## 7 1_95 52 1 52 0.639 0.000000567
## # ℹ 16 more variables: ts_linearity <dbl>, ts_curvature <dbl>, ts_e_acf1 <dbl>,
## # ts_e_acf10 <dbl>, ts_seasonal_strength <dbl>, ts_peak <dbl>,
## # ts_trough <dbl>, ts_entropy <dbl>, ts_x_acf1 <dbl>, ts_x_acf10 <dbl>,
## # ts_diff1_acf1 <dbl>, ts_diff1_acf10 <dbl>, ts_diff2_acf1 <dbl>,
## # ts_diff2_acf10 <dbl>, ts_seas_acf1 <dbl>, ts_my_mean <dbl>Clustering with K-Means
We can quickly add cluster assignments with the kmeans()
function and some tidyverse data wrangling.
set.seed(123)
cluster_tbl <- tibble(
cluster = tsfeature_tbl %>%
select(-id) %>%
as.matrix() %>%
kmeans(centers = 3, nstart = 100) %>%
pluck("cluster")
) %>%
bind_cols(
tsfeature_tbl
)
cluster_tbl## # A tibble: 7 × 23
## cluster id ts_frequency ts_nperiods ts_seasonal_period ts_trend ts_spike
## <int> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 2 1_1 52 1 52 0.000670 0.0000280
## 2 2 1_3 52 1 52 0.0614 0.00000987
## 3 2 1_8 52 1 52 0.756 0.00000195
## 4 1 1_13 52 1 52 0.354 0.00000475
## 5 3 1_38 52 1 52 0.425 0.0000179
## 6 3 1_93 52 1 52 0.791 0.000000754
## 7 1 1_95 52 1 52 0.639 0.000000567
## # ℹ 16 more variables: ts_linearity <dbl>, ts_curvature <dbl>, ts_e_acf1 <dbl>,
## # ts_e_acf10 <dbl>, ts_seasonal_strength <dbl>, ts_peak <dbl>,
## # ts_trough <dbl>, ts_entropy <dbl>, ts_x_acf1 <dbl>, ts_x_acf10 <dbl>,
## # ts_diff1_acf1 <dbl>, ts_diff1_acf10 <dbl>, ts_diff2_acf1 <dbl>,
## # ts_diff2_acf10 <dbl>, ts_seas_acf1 <dbl>, ts_my_mean <dbl>Visualize the Cluster Assignments
Finally, we can visualize the cluster assignments by joining the
cluster_tbl with the original
walmart_sales_weekly and then plotting with
plot_time_series().
cluster_tbl %>%
select(cluster, id) %>%
right_join(walmart_sales_weekly, by = "id") %>%
group_by(id) %>%
plot_time_series(
Date, Weekly_Sales,
.color_var = cluster,
.facet_ncol = 2,
.interactive = FALSE
)
Learning More
My Talk on High-Performance Time Series Forecasting
Time series is changing. Businesses now need 10,000+ time series forecasts every day. This is what I call a High-Performance Time Series Forecasting System (HPTSF) - Accurate, Robust, and Scalable Forecasting.
High-Performance Forecasting Systems will save companies MILLIONS of dollars. Imagine what will happen to your career if you can provide your organization a “High-Performance Time Series Forecasting System” (HPTSF System).
I teach how to build a HPTFS System in my High-Performance Time Series Forecasting Course. If interested in learning Scalable High-Performance Forecasting Strategies then take my course. You will learn:
- Time Series Machine Learning (cutting-edge) with
Modeltime- 30+ Models (Prophet, ARIMA, XGBoost, Random Forest, & many more) - NEW - Deep Learning with
GluonTS(Competition Winners) - Time Series Preprocessing, Noise Reduction, & Anomaly Detection
- Feature engineering using lagged variables & external regressors
- Hyperparameter Tuning
- Time series cross-validation
- Ensembling Multiple Machine Learning & Univariate Modeling Techniques (Competition Winner)
- Scalable Forecasting - Forecast 1000+ time series in parallel
- and more.