Student guide: Data health and Data

A log is only as good as the data behind it. The Data health tab lists what the import found in the workbook: problems in the site’s data, and what the app assumed. The Data tab shows every sheet as the app read it. Look at both whenever a log surprises you.

Nothing in your data is changed. The app reports what looks wrong, says why it matters and what to do, and leaves the fixing to the database, where it belongs.

The Data health tab

1. Reading the findings

The counts at the top, also on the load line under the header, sort the findings into three kinds:

Kind	Means	Example
To fix (red)	The log can’t be drawn right until the data are changed.	A sheet is missing; an interval’s FROM is deeper than its TO.
To check (amber)	Probably a problem in the site’s data.	Two overlapping intervals; a survey that jumps 30° between stations.
For information (grey)	What the app found or assumed.	Which categories exist; which sign the dips use.

The table lists them worst first. Each row has the Area (sheets, collar, survey, geology, assays, QC, coordinates), the Finding, how many Rows it affects, a Detail, Why it matters and What to do. Select a row to see up to 200 of the rows behind it.

Example: the demo has 0 to fix, 9 to check and 28 for information. Among the nine to check: “Category names that look like the same category” (Lithology / Litholoigy, a typing slip at the site), “Grid name and projection zone disagree” (a grid called UTM50S whose projection says UTM zone 48N), and “Recovery or RQD longer than its run” (4 runs).

2. The checks, and why they exist

Sheets and columns

The workbook should have eleven sheets: Collar, HoleDetails, Survey, Geology, SampleIntervals, Assays, QCDuplicates, QCStandardsBlanks, Grids, GridPickup and HoleCoords (see Input format). They are recognised by their columns, so a renamed sheet is still found, and the app says so.

Finding	Why it matters
View missing; required column missing	Part of the log can’t be drawn.
Number or date stored in a text column; text in a number column	Usually paste damage: a spreadsheet changed the cell’s type, for example turning hole 000123 into 123.
Date not readable; date in the future	A typing slip or a wrong date format.

Holes, depths and coverage

Finding	Why it matters
Duplicate hole in the collar sheet	Two collars for one hole: which is right?
Hole not in the collar sheet	Data for a hole the workbook doesn’t define. If removing leading zeros makes it match, the IDs were damaged by a spreadsheet.
Interval or station deeper than the hole	The hole was deepened and the collar depth not updated, or a depth was typed wrongly.
FROM greater than TO	An interval upside down: it can’t be drawn.
Hole with no depth	The log runs to the deepest data instead.

Survey

The survey gives the hole’s path. Errors here put every depth in the wrong place in space.

Finding	Why it matters
Stored grid azimuth looks 180° out	The stored azimuth is the back-bearing of the logged one: the hole points the wrong way.
Logged and calculated azimuth disagree; azimuth not on the collar’s grid	Azimuths must be on the same grid as the collar.
Dip sign convention inferred; dip changes sign within a hole	Some sites record down as negative, others as positive. A hole that changes sign has a typing error.
Two stations at the same depth; sudden change between stations (azimuth > 20° or dip > 10°)	A repeated or mistyped station bends the hole where it didn’t bend.

Geology

Finding	Why it matters
Overlapping intervals for one attribute and priority	Two rock types at one depth: the log can’t say which.
Category names that look like the same category	For example “Litholoigy” beside “Lithology”: a typing slip splits one category in two.
Angle outside its possible range	Alpha must be 0 to 90 and beta 0 to 360.
Recovery or RQD longer than its run; percentage outside 0-100	Impossible values, usually a wrong field or a typing slip.
Code not in the site’s lookup list	A code no one defined, so it has no description.

Assays and QC

Finding	Why it matters
Sample and element with only pending or failed results	These are shaded on the log and don’t count as grade in intercepts.
Below-detection results; negative values	How the site stores results below the detection limit. Only results marked < are treated as below detection.
Element reported in more than one unit	The app converts them; check the unit on each column.
Assay with no rank; several methods sharing a rank	The app can’t tell which method the site prefers, so the latest load wins.
Sample depths differ between Assays and SampleIntervals	The same sample in two places.
Duplicates, standards and blanks	Summaries of the QC samples sent with the assays. Standards with no certified value can’t be judged.

Why QC matters: a laboratory checks its own work with duplicates (the same sample assayed twice), standards (material of known grade) and blanks (material with none). A batch that fails these checks is not trusted, and the log shows its results shaded.

Grids and coordinates

Finding	Why it matters
Grid name and projection zone disagree	A grid called “zone 50” with a projection in zone 48 puts holes hundreds of kilometres out.
Local grid with no pickup points	A mine grid can’t be converted without points known in both grids.
Collar more than 10 m from another of its coordinate sets	Often planned against surveyed positions, which is expected; otherwise one is wrong.

3. The Data tab

Sheet chooses any sheet of the workbook. The table shows it as the app read it: numbers as numbers, dates as dates, IDs as text. Type in a box above a column to filter it.

Why: when a finding or a log looks wrong, find the row here and check the value at its source. For example, filter the Geology sheet by HOLEID and ATTRIBUTE to see every interval behind a lithology column.

Exercises

1. Read a finding. In the demo, find “Category names that look like the same category”. Select it: which two names are they? What would you ask the database administrator to do?
2. Overlaps. Find “Overlapping intervals for one attribute and priority”. Which hole is it? Open that hole on the Log tab and find the overlap. What does the log show there?
3. RQD. Find “Recovery or RQD longer than its run”. Look at one of its rows on the Data tab (Geology sheet). Is it more likely a typing slip or a wrong field?
4. Your own workbook. Load a workbook from your own project. How many findings are there to fix? Pick one to check and follow its “What to do”.