A chart labeled Miami building permits can describe very different things depending on its source. It might count permit records issued by the City of Miami, housing units authorized in a Census series or activity across a broader metropolitan area. Those are not interchangeable measures. Before interpreting an increase as future apartment supply, a reader should establish exactly what was counted, where it occurred and which stage of the construction process the record represents.
The City of Miami’s Data Explorer provides a local starting point. It links to a building permits data story and an open dataset described as covering permits issued by the city’s Building Department since 2014. The Census Building Permits Survey provides another framework with its own definitions. This guide explains how to reconcile those source types without inventing a trend from records that have not been classified consistently.
Begin with a claim you can actually test
Write the proposed conclusion before downloading anything. “The number of city permit records increased” is different from “more homes were authorized” and very different from “more apartments will be available next summer.” The first requires a defined record count. The second requires a housing unit measure. The third requires evidence about completion and availability that an issuance count alone does not provide.
For a local rental research question, a useful starting claim is narrower: how much of the identified permit activity concerns new residential construction rather than other work? That directs attention to the dataset’s fields and classifications. It also avoids assuming that every record represents an additional home. A roof repair, an interior alteration and a new building can all be relevant construction activity while contributing differently to a housing supply analysis.
Do not use a headline to decide the classification after the fact. Establish inclusion rules first, then apply them to the records. If the dataset cannot support the distinction needed, say so. An honest limitation is more informative than a precise looking total built from an unsupported interpretation of a status or description field.
Preserve the City of Miami boundary
The local Data Explorer describes City of Miami data. Keep that boundary in the title and notes of any result derived from it. Do not rename it Miami Dade County or the Miami metropolitan area because those labels are more familiar to readers. A broader region can contain activity outside the city, and a city dataset cannot by itself describe the full regional construction pipeline.
For a Census comparison, record the geography exactly as supplied by the selected file. Store the identifier as well as the name. If the desired geography is a permit issuing place, read its coverage notes. Census explains that such a place can represent different jurisdictional arrangements, and footnotes can clarify coverage that is not apparent from the label alone.
A good worksheet contains a boundary field for each source, not one shared field at the top. This makes mismatches visible before the figures are combined. If the sources cover different areas, they may still be discussed as separate context, but they should not be subtracted from one another as though they measure the same territory.
Read record dates as events
The city describes its open permit information as including application data, status and review timeline attributes. A date field therefore needs an explicit interpretation. Application date, issue date and later record update date can place the same project into different annual groups. Select the date that matches the question and retain the field name in the methodology.
If measuring permits issued during a calendar year, use the appropriate issue event rather than the date the file was downloaded. If measuring applications submitted, identify that as a different series. Avoid mixing events because one record lacks the preferred date. Instead, count missing dates separately and explain how they were handled.
A fictional project might be submitted in December and issued in February. Grouping by submission puts it in one year; grouping by issuance puts it in the next. Neither calculation is automatically wrong, but they answer different questions. The error occurs when a chart changes between those rules without telling the reader.
Identify what one row represents
Inspect the dataset documentation and a small set of records before calculating a total. Determine whether the chosen identifier refers to a permit, process, project or another record type. The city’s permit history guidance distinguishes a process number associated with an application from a permit number issued when a permit is pulled. Preserve those labels instead of renaming every identifier project ID.
A single property can have several related permits. A single construction project can involve more than one type of work. Conversely, a permit can concern a building with many housing units. Counting rows therefore cannot automatically produce a number of new homes. The analyst must identify a verified unit measure or explicitly limit the result to permit records.
Check for duplicate identifiers in the downloaded file, but do not delete records simply because their addresses match. The same address can legitimately appear more than once. A duplicate test should examine the dataset’s actual key and event structure. Save a log of any removed duplicate records so another reviewer can understand the decision.
Understand the Census measure separately
Census defines authorization as local approval through building or zoning permits and distinguishes authorized units from starts and completions. Its housing unit statistics have exclusions, including units created within existing structures. That matters when comparing the survey with a local dataset containing renovation activity. A local conversion or alteration record may not belong in the same statistical universe as the selected Census series.
The Census definitions also distinguish reported data from estimates that include imputation. Retain that label with a downloaded figure. An estimate with imputation is not simply a list of observed permit rows, and a missing local report should not be silently converted into zero. The source’s treatment of incomplete reporting belongs in the methodology.
Keep this national definition context separate from a claim about Miami’s actual numbers. Definitions explain how to interpret a series; they do not establish that local activity increased or decreased. This article publishes no unverified five year trend and makes no forecast about rents or completion dates.
A fictional reconciliation exercise
Imagine a fictional city export containing 120 records for a chosen year. The analyst identifies 70 as repairs, 20 as alterations, 10 as new residential work and 20 as categories that remain unclear. These are invented training values, not City of Miami observations. The correct first result is a classification table showing those categories and the unresolved group, not a claim that 120 homes were added.
Suppose the ten new residential records refer to six projects after related records are reviewed. Even then, six projects do not establish six housing units. One may contain multiple units. The analyst needs a verified unit field or project documents before making a housing authorization total. If that evidence is unavailable, the report should retain the project or permit measure and state the limitation.
Now suppose an unrelated Census table shows a fictional total of 90 authorized units for a broader geography. It would be invalid to conclude that the city export missed 84 homes by subtracting six from 90. The measures and geography differ. The reconciliation task is to align definitions first, not to force two unequal numbers to match.
Make a reproducible extraction note
Save the source page, download date, file name, selected geography, date field, status filter and classification rule. Record whether the export was complete or limited by a dashboard filter. If a tool offers both a chart and downloadable records, check whether the same selection is applied to each. A screenshot of a chart is not enough to reconstruct a hidden filter.
Keep the original download unchanged and perform cleaning on a copy. Use descriptive column names in the working file, but preserve the source column names in a mapping note. This makes it possible to reproduce the analysis after a website update or to identify why a colleague obtains a different total.
For a future multiyear table, use one row per year and columns for the chosen measure, coverage note, missing data treatment and revision status. Do not fill gaps by carrying forward the previous year without an explicitly justified method. A blank with an explanation is preferable to a manufactured trend line.
Connect aggregate data to a specific building cautiously
A regional authorization series cannot tell a renter whether a particular advertised apartment is ready. For that question, use the city’s property specific permit history route and the actual project identifiers. Ask which records relate to the advertised work and whether additional documentation is needed to understand status. Do not infer present condition from an aggregate construction chart.
Similarly, one delayed project does not prove that the aggregate series is wrong. The series and the property file describe different levels of evidence. A useful report can present both, but it should explain the connection rather than treating a single example as representative of every project in the city.
Publish the narrow conclusion
A defensible conclusion names the geography, period, measure and principal limitation. It might describe a count of issued city permit records in selected categories, or a Census total of authorized units under its definitions. It should avoid describing either as completed rental homes unless completion evidence is actually available.
For Miami readers, the value of this workflow is the separation of local administrative activity from housing supply claims. The City’s data tools reveal records worth investigating; Census definitions clarify a statistical measure. Used together with explicit boundaries and a reproducible method, they support careful research without turning permits into a promise about future rents or move in dates.