Summary

Check that you can inspect a text file, import it with explicit settings, validate its contents, and choose an appropriate export format.

Before you start: selecting table columns; project folders and relative paths.
Setup: Use Data/quakes.csv from the import lesson. Practice writes new files under Output/.
Your check: Adapt export settings and verify the saved values and structure. Practise in the unit summary.

What you learned

  1. Inspect before importing. Identify the source, reporting period, units, metadata lines, header, delimiter, decimal mark, encoding, and missing-value symbols. A .csv extension alone does not specify these settings.
  2. Make the import reproducible. Keep the raw download and use code that records the settings. read.table() and its CSV variants read delimited text; an Excel workbook requires a suitable reader or a documented export to text first.
  3. Validate the result. Use head(), str(), and dim(). Check column names and types, row counts, missing values, plausible ranges, and the meaning of identifiers.
  4. Clean with a reason. Preserve identifiers as text when leading zeros matter. Distinguish missing values from zero and identify regional levels and overlapping categories before calculating totals.
  5. Export and check again. Reimport an exported file to confirm that values and important identifiers survived.

A short recap in R

Use the quakes.csv download from the import lesson, saved as Data/quakes.csv inside your RStudio project. Start from the project root. This particular teaching CSV has four metadata lines, a header, semicolon separators, and decimal commas.

quakes_data <- read.table("Data/quakes.csv",
                          skip = 4, header = TRUE, sep = ";", dec = ",")
dim(quakes_data)
head(quakes_data)
str(quakes_data)

The result should have 1,000 rows and five numeric columns: latitude (lat), longitude (long), depth in kilometers (depth), magnitude (mag), and number of reporting stations (stations). These are checks of this file, not defaults for another dataset.

Settings and formats to remember

Setting or format Why it matters
header, skip, sep, dec Identify the table and correctly separate names and values.
fileEncoding, na.strings Decode text correctly and recognize documented missing-value markers.
colClasses Specify column types, for example character for region identifiers.
CSV with write.table() or write.csv() Exchanges tabular text; choose separators/decimal marks and usually row.names = FALSE. R-specific classes are not automatically preserved.
RDS with saveRDS() and readRDS() Saves and restores one R object, including its structure and classes.

The Destatis example demonstrates why an official source still needs careful preprocessing: a German statistical export includes metadata, coded names, decimal commas, blanks, and several geographic levels. The dictionary and source notes are part of understanding the data, not optional decoration. Do not add Germany to its own states or sum parent categories together with their subcategories.

For batch processing, first test the complete import and validation on one file. Then identify files explicitly with list.files() and build paths with file.path(). The following unit explains the loops and functions needed to repeat those steps reliably.

Try it yourself

Use the supplied Quakes CSV at Data/quakes.csv with the project root as working directory. Predict which import settings it needs, then import and inspect its dimensions/types. Export the first three rows to Output/quakes-practice.csv using semicolons and a decimal point. Read the export back and compare it with those rows. Also save and restore the same rows as RDS.

Use these new output names for practice; keep the input unchanged. Expect a 1,000-by-5 import, a 3-by-5 export and two successful round-trip comparisons. The export starts with its header, not the four input metadata lines.

Hint

The input settings are shown in the recap above. Use row.names = FALSE when writing the CSV; choose the reader settings from your export format. Create Output/ if it does not yet exist.

Compare a possible solution 
quakes_practice <- read.table("Data/quakes.csv", skip = 4,
                              header = TRUE, sep = ";", dec = ",")
dim(quakes_practice)
str(quakes_practice)
first_three <- quakes_practice[1:3, ]
dir.create("Output", showWarnings = FALSE)
write.table(first_three, "Output/quakes-practice.csv", sep = ";",
            dec = ".", row.names = FALSE)
restored_csv <- read.table("Output/quakes-practice.csv", header = TRUE,
                           sep = ";", dec = ".")
identical(first_three, restored_csv)
saveRDS(first_three, "Output/quakes-practice.rds")
identical(first_three, readRDS("Output/quakes-practice.rds"))

Both comparisons return TRUE for these numeric data. This does not imply that CSV preserves every R class or attribute. To practise the official-data decisions, explain why the Destatis example keeps identifiers as text, excludes the national aggregate from state totals and separates parent categories from their parts.

For the batch preview, describe the one-file operation before returning after Unit 07.

Check your batch-processing explanation

For each selected input file, construct its path, import with verified settings, check its structure and meaning, then write a separately named output. Test this whole operation on one file first. A loop repeats the operation; it does not make incompatible files comparable or validate them automatically. Keep the original inputs unchanged.

Ready to continue: you can choose settings from a file, check the imported values and complete a documented round trip.

Check your understanding

Try these questions before opening the suggested answers.

  1. A CSV imports as one column. Which setting would you inspect first, and how would you check the raw file?
  2. Why might a region code such as "01" require character storage, while an area in hectares requires numeric storage?
  3. Why can an import finish without errors and still give an unsuitable dataset for analysis?
Suggested answers
  1. Inspect the field separator and the raw header/first data lines in a text editor. Choose sep from the file, rather than from its extension or language; check skip and header as well.
  2. A region code identifies a place and may require leading zeros. Area in hectares is a quantity on which calculations make sense, so its values should be numeric with documented units.
  3. Import success means the file could be parsed. Wrong column types, decimal marks, missing-value rules, duplicate keys or mixed geographic levels can still make the result unsuitable. Check structure, dimensions and meaning against the source.

Revisit import and export and batch processing if needed.

Test what you’ve learned so far!

This repeats the path and data-import quiz from the import lesson. Try it again without looking at the examples; use its feedback to choose what to revisit.

Updated: