Statistical Histogram
A distribution plot that partitions numerical observations into bins and represents counts or probability density by adjacent rectangles.
A statistical histogram partitions numerical observations into adjoining intervals. Each rectangle summarizes the observations assigned to its interval. Unlike a bar chart of categories, the horizontal axis carries a quantitative scale and the bins occupy contiguous ranges.
Counts, density and area
Let bin i contain f_i of n observations and have width h_i. Plotting f_i as height is a count histogram. When widths differ, those heights cannot be compared as though areas were proportional to frequency. A density histogram uses height f_i/(n × h_i); the rectangle has area f_i/n, and the total area is 1. With equal bin widths, count and density histograms differ only by vertical scaling.
Suppose 20 of 100 measurements fall in an interval of width 2. Its relative frequency is 0.20 and density height is 0.10. For the same count in an interval of width 5, the area remains 0.20 but the height becomes 0.04.
Boundary and resolution choices
Changing the bin width or the alignment of edges can alter the apparent number of modes, gaps and tails without changing any observation. Many numerical libraries use half-open intervals [a,b) and include the right edge of the last bin; this convention should be verified rather than assumed. Constant-valued data, missing measurements, zero range and weighted observations require explicit handling.
Bin count and width can be estimated with Sturges, Freedman–Diaconis or Scott's rule. A frequency table reports counts; an ECDF reports cumulative proportions without choosing bins. The observability histogram and database histogram are distinct technical uses of the same word.