Representation is a reasoning choice. A line graph is useful for continuous change in an ordered variable such as time; a scatter plot shows association between paired numerical variables; a bar chart compares categories; a table preserves exact values. Statistics compress a dataset but can hide distribution or anomalies.
Axes need variable names and units. Scales should reveal the data honestly. A mathematical relationship or model is useful only when supported by the observed pattern and stated domain; do not force a trend line through data merely because software offers one.