[ZEPPELIN-3806] Misleading x-axis visualisation when data has gaps

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 0.8.0
• Component/s: GUI

Environment

OSX, Chrome 69.0.3497.100, Zeppelin 0.8.0

• Assignee: Unassigned
• Reporter: Mikko Lipasti
• Created: 02/Oct/18 11:45
• Updated: 02/Oct/18 11:49

Description

Given time-series data with gaps in the data (e.g. per-hour counts, some hours have no data), when plotted (especially when using scatterplot), gap in the data should be clearly visible as a gap at the missing hour.

Instead, Zeppelin visualisation completely ignores the missing data.

%spark
import java.sql.Timestamp
val d1 = (Timestamp.valueOf("2018-10-02 11:00:00"), 10)
val d2 = (Timestamp.valueOf("2018-10-02 12:00:00"), 11)
val d3 = (Timestamp.valueOf("2018-10-02 14:00:00"), 10)
val d4 = (Timestamp.valueOf("2018-10-02 15:00:00"), 9)

val data = Seq(d1, d2, d3, d4).toDF("timestamp", "count")
z.show(data)

The distance on the x-axis is incorrect. With scatterplot one would expect that x-axis is linear but it is not: what looks like equal distances vertically can actually be different, based on data.