Data Analysis of Impact of Each NBA Position

Based on 2020-2021 NBA Season

Michael Fischgrund (mf788), Kyrah Barnett (klb332), Dubby Ugha (nfu2), Isabel Pangestu (isp24)

5/4/23

Introduce the data

  • Our group loves watching basketball, especially the NBA (National Basketball Association).

  • We were really interested in seeing which position (ex. point guard, center) is the best on offense.

  • We wanted to create one statistic that pools together the four most important statistics (points, assists, offensive rebounds, turnovers). We call this the ‘positional impact rating’.

  • Research question: Is there a difference between the positional impact ratings of NBA positions?

Highlights from EDA

  • Shooting guards have the highest points per game category by a noticeable amount, while small forwards have the least points per game.

  • Point guards average the most assists per game while centers average the least, and centers average the highest field goal percentage while shooting guards average the lowest.

# A tibble: 5 × 9
  Pos   mean_ppg mean_apg mean_rpg mean_efg mean_fgp mean_threep_pct mean_ftpct
  <chr>    <dbl>    <dbl>    <dbl>    <dbl>    <dbl>           <dbl>      <dbl>
1 C         8.37     1.29    1.64     0.546    0.519           0.243      0.695
2 PF        7.56     1.50    0.842    0.508    0.442           0.314      0.732
3 PG        9.86     3.56    0.452    0.485    0.416           0.339      0.797
4 SF        7.81     1.42    0.660    0.504    0.426           0.330      0.754
5 SG        9.39     1.97    0.452    0.498    0.412           0.334      0.782
# ℹ 1 more variable: mean_tov <dbl>

Visualizations

As shown below, point guards have the highest position impact rating.

Inference/modeling/other analysis

  • We did a hypothesis test to compare the means of the positional impact ratings between point guards and shooting guards.

  • The observed difference in means was 1.703048.

  • The p-value obtained from the test was 0.072. We fail to reject the null hypothesis, concluding that there is no significant difference in the means.

[1] 1.703048
# A tibble: 1 × 1
  p_value
    <dbl>
1   0.062

Conclusions + future work

  • The p-value obtained from the test was 0.072. We fail to reject the null hypothesis, concluding that there is no significant difference.

  • From this we are able conclude that the point guard and shooting guard positions are statistically the same in terms of offensive impact.

  • In terms of our future work, we want to compare other positions to each other, and take into account even more offensive statistics so we can have a more holistic and accurate ‘positional impact rating’.