Beginning Anomaly Detection Using

 art_daily_jumpsdown

This data set has mixture of normal data and anomalies. As you can see below, the time 

series has values at different timestamps.

Using visualization, you can plot the time series now. You convert the timestamp to 

datetime for this work and also drop the timestamp column. As shown below, the time 

series shows the datatime vs. the value column.

Dataset: art_daily_jumpsdown.csv

Figure 6-42.  A graph showing anomalies

Chapter 6   Long Short-term memory modeLS 

---

247

Figure 

6-43

 shows the code to generate a graph showing the time series.

Figure 6-43.  A graph showing the time series

Let’s add the anomaly column to the original dataframe and prepare a new 

dataframe. Using visualization, you can plot the new time series now. As shown below, 

the time series shows the datatime vs. the value column. Normal data points are shown 

in green and anomalies are shown in red. Figure 

6-44

 shows the code to generate a graph 

showing anomalies.

Chapter 6   Long Short-term memory modeLS 

---

248

Since this data set has some noise or anomalies, there are anomalies (datapoints in 

RED) shown and everything else that is normal is green.

Next, let’s examine another dataset which is different from the current dataset. You 

will build a LSTM model and see if there are anomalies or not.