Graphics Reference
In-Depth Information
Figure
.
.
[his figure also appears in the color insert.] Relationships among functional objects: the
let panel shows a
-D scatterplot of opening bid (x), day of the auction (y)andprice(z). he right
panel shows a smoother version of the price surface, obtained using a Nadaraya-Watson smoother. In
that plot, the x-axis is the opening bid and the y-axis is the day of the auction
Fig.
.
shows a smoother image of the price surface obtained using a Nadaraya-
Watson smoother. he three-way relationship between opening bid, price and time
is now much easier to see.
Visualizing Functional and Cross-sectional Information
5.4.3
As illustrated above, it is more challenging to visualize functional data than classi-
cal data. he visualization process is oten further complicated by the coupling of
functional observations with cross-sectional attribute data. For example, online auc-
tion data include not only the bid history (i.e., the times and sizes of bids), but also
auction-specific attributes corresponding to auction design (e.g., length of the auc-
tion, magnitude of the opening bid, use of a secret reserve price, use of the “Buy It
Now" option, etc.), bidder characteristics (e.g., bidder IDs and ratings), seller char-
acteristics(e.g.,sellerIDandrating,sellerlocation,whetherornotasellerisa“Pow-
erseller,"etc.),andproductcharacteristics(e.g.,productcategory,productqualityand
quantity, product description, etc.). All of these characteristics correspond to cross-
sectional information in that they do not change during the auction. he coupling
of time series with cross-sectional information is important because the relationship
between the two could be the main aim or at least one of the aims of the analysis.
