KaneAI - World’s First E2E Software Testing Agent.Book a Demo 
How to use touchDown method in fMBT
Best Python code snippet using fMBT_python
touchdown.py
Source:touchdown.py 
1# -*- coding: utf-8 -*-2"""3Created on Thu Mar 10 21:28:51 20164@author: Tobias Jachowski5"""6import matplotlib.pyplot as plt7import numpy as np8from matplotlib.widgets import Cursor9from scipy.optimize import fsolve10from pyoti.modification.modification import Modification, GraphicalMod11from pyoti import traces as tc12class ITouchdown(GraphicalMod):13 """14 Subclass of Touchdown that provides graphical interfaces to adjust the fit15 parameters.16 """17 def __init__(self, **kwargs):18 super().__init__(**kwargs)19 # Define attributes needed for the graphical fitting of the touchdown20 # a cursor on the axes to show upper fitting border and left/right21 # (middle) border22 self._cursor = None23 self._lines = {}24 self._ax = None25 self.supertitle = None26 def _figure(self):27 """28 Initialize and show an interactive plot for fitting the touchdown.29 Adjust the fit parameters interactively by displaying the fitted30 touchdown. The plot of the fitting is stored in self.figure.31 """32 # create new figure and axes for fitting the touchdown33 figure, ax = plt.subplots(1, 1, sharex=True, sharey=True)34 self._ax = ax35 # create lines to plot the data for the fit and for displaying fitting36 # result37 self._lines['orig_data'] = ax.plot([0], [0], tc.color('psdZ'))[0]38 self._lines['binned_data'] = ax.plot([0], [0], 'b.')[0]39 # cursor to display the "boundaries" for fitdata40 self._cursor = Cursor(ax, useblit=True, color='red', lw=1)41 # initialize all lines for the plot that represent the fit paramters42 # extrapolation of fit left43 self._lines['ext_left'] = ax.plot([0, 0], [0, 0], 'c', lw=1)[0]44 # extrapolation of fit right45 self._lines['ext_right'] = ax.plot([0, 0], [0, 0], 'c', lw=1)[0]46 # fit left47 self._lines['fit_left'] = ax.plot([0, 0], [0, 0], 'r', lw=3)[0]48 # fit right49 self._lines['fit_right'] = ax.plot([0, 0], [0, 0], 'g', lw=3)[0]50 # touchdown position51 self._lines['touchdown'] = ax.axvline(x=0.0, color='y', lw=2)52 # touchdown of bead, where positionZ equal to 0.053 self._lines['touchdown_calib'] = ax.axvline(x=0.0, color='k', lw=2,54 dashes=(10, 10))55 ax.set_xlabel(tc.label('positionZ'))56 ax.set_ylabel(tc.label('psdZ'))57 ax.ticklabel_format(useOffset=False)58 ax.grid(True)59 self.supertitle = figure.suptitle("Fit touchdown determination (move "60 "to upper and right/left limit, "61 "left/right click, repeat ...)")62 # method to update plot and perform the fit63 figure.canvas.mpl_connect('button_release_event',64 self._update_touchdown)65 return figure66 def _update_fig(self, data_changed=True, **kwargs):67 """68 Updates the interactive fitting plot: Set values of fit parameters,69 touchdown position and fitted lines70 """71 if data_changed:72 # data, [:,0] is positionZ, [:,1] is psdZ73 data = self.modification._get_data_based(74 traces=['positionZ', 'psdZ'], decimate=True, copy=False)75 self.modification.validate_fit_params(data=data)76 bin_means = self.modification.bin_means77 minx = data[:, 0].min() - 0.05e-6 # m78 maxx = data[:, 0].max() + 0.05e-6 # m79 miny = data[:, 1].min() - 0.0580 maxy = data[:, 1].max() + 0.0581 self._ax.set_xlim([minx * 1e6, maxx * 1e6]) # µm82 self._ax.set_ylim([miny, maxy])83 self._lines['orig_data'].set_data(data[:, 0] * 1e6, data[:, 1])84 self._lines['binned_data'].set_data(bin_means[:, 0] * 1e6,85 bin_means[:, 1])86 # get left and right datapoint means according to fit borders for87 # fitting88 bin_means = self.modification.bin_means89 left_means = self.modification.left_means90 right_means = self.modification.right_means91 self._lines['ext_left'].set_data(bin_means[:, 0] * 1e6,92 self.get_pv_left(bin_means[:, 0]))93 self._lines['ext_right'].set_data(bin_means[:, 0] * 1e6,94 self.get_pv_right(bin_means[:, 0]))95 self._lines['fit_left'].set_data(left_means[:, 0] * 1e6,96 self.get_pv_left(left_means[:, 0]))97 self._lines['fit_right'].set_data(right_means[:, 0] * 1e6,98 self.get_pv_right(right_means[:, 0]))99 self._lines['touchdown'].set_xdata([self.modification.touchdown * 1e6,100 self.modification.touchdown * 1e6])101 self._lines['touchdown_calib'].set_xdata(102 [self.modification._calibration.touchdown * 1e6,103 self.modification._calibration.touchdown * 1e6])104 def _pre_close_fig(self):105 # Store touchdown fit plot for documentation106 self.supertitle.set_text("Touchdown determined by polynomial fits")107 self._cursor.visible = False108 # update visibility status of self._cursor before removing reference109 # to it110 self._figure_canvas_draw()111 # Release unused memory112 self._cursor = None113 self._ax = None114 self._lines.clear()115 self.supertitle = None116 def _update_touchdown(self, event):117 """118 Gets called by events of the canvas object of the figure119 """120 self._set_fit_params(event)121 touchdown = self.modification.fit_touchdown()122 self.modification._set_touchdown_leave_auto(touchdown)123 self.update_fig(data_changed=False)124 def _set_fit_params(self, event):125 """126 Set the parameters for the fitting routine.127 """128 if event.button == 1: # left mouse button -> adjust left fitdata129 # left border for fitdata130 self.modification.left = self._cursor.linev.get_xdata()[0] * 1e-6 # m131 # max value (psdZ) for fitdata132 self.modification.left_upper = self._cursor.lineh.get_ydata()[0]133 else: # right (or middle) mouse button -> adjust right fitdata134 # right border for fitdata135 self.modification.right = self._cursor.linev.get_xdata()[0] * 1e-6 # m136 # max value (psdZ) for fitdata137 self.modification.right_upper = self._cursor.lineh.get_ydata()[0]138 def get_pv_left(self, x):139 return np.polyval(self.modification.pf_left, x)140 def get_pv_right(self, x):141 return np.polyval(self.modification.pf_right, x)142class Touchdown(Modification):143 """144 Determine the approximate touchdown145 The displacement sensitivity and stiffness of a calibration are always146 relative to and given absolute at the touchdown of the bead on the surface147 The positionZ is touchdown corrected by this modification.148 """149 GRAPHICALMOD = ITouchdown150 def __init__(self, fit_touchdown=True, **kwargs):151 traces_apply = ['positionZ']152 # the touchdown position determined by calibration and fitting153 super().__init__(automatic_switch=True, datapoints=2500,154 traces_apply=traces_apply, **kwargs)155 # register a widget, giving a key, a function called upon change156 self.add_iattribute('touchdown', description="Touchdown (m)",157 value=0.0)158 # Initially set touchdown to touchdown of calibration, if touchdown159 # should not be automatically fitted. This will automatically switch of160 # the automatic fitting.161 if not fit_touchdown:162 self.touchdown = self._calibration.touchdown163 # register a button to set the touchdown from calibration.touchdown164 self.add_iattribute('set_touchdown_calib',165 description="Set touchdown from calibration",166 callback_functions=[167 self._set_touchdown_from_calibration,168 self.evaluate])169 # Define attributes needed for the fitting of the touchdown170 self.left_upper = None # max value (psdZ) for fitdata171 self.right_upper = None # max value (psdZ) for fitdata172 self.left = None # right/left border for fitdata173 self.right = None # right/left border for fitdata174 def _recalculate(self):175 # calculate data for fitting176 self.validate_fit_params()177 # fit the touchdown178 touchdown = self.fit_touchdown()179 self._set_touchdown_leave_auto(touchdown)180 def _modify(self, data, samples, data_traces, data_index, mod_index):181 data[:, data_index] -= self.touchdown182 return data183 def validate_fit_params(self, data=None):184 # self.init_fit_parameters()185 if data is None:186 data = self._get_data_based(traces=['positionZ', 'psdZ'],187 decimate=True, copy=False)188 positionzmin = data[:, 0].min()189 positionzmax = data[:, 0].max()190 middle = positionzmin + (positionzmax - positionzmin) / 2191 upper = data[:, 1].max()192 # First initialisation of fit boundaries193 if self.left is None or self.right_upper is None:194 self.left_upper = upper # max value (psdZ) for fitdata195 self.left = middle # right/left border for fitdata196 if self.right is None or self.right_upper is None:197 self.right = middle # right/left border for fitdata198 self.right_upper = upper # max value (psdZ) for fitdata199 bin_means, bin_width = self.calculate_bin_means(data=data)200 self.bin_means = bin_means201 if not np.any(self.left_means_idx):202 # first, change only the upper boundary and leave the left one203 # usually solves problems, when psdZ was changed by an offset204 self.left_upper = upper205 if not np.any(self.left_means_idx):206 # still no means, change left boundary, too207 self.left = middle208 if not np.any(self.right_means_idx):209 # first, change only the upper boundary and leave the right one210 # usually solves problems, when psdZ was changed by an offset211 self.right_upper = upper212 if not np.any(self.right_means_idx):213 # still no means, change right boundary, too214 self.right = middle215 def fit_touchdown(self):216 """217 Performs the fit.218 In essence, it fits 2 polynomials of 2nd degree to the left (bead219 on surface) and to the right (free bead) portion of the psdZ data220 The intercept of these two polynomials determine the touchdown of the221 bead with the surface.222 """223 if self.left_means.shape[0] > 0 and self.right_means.shape[0] > 0:224 def f_left(x):225 return np.polyval(self.pf_left, x)226 def f_right(x):227 return np.polyval(self.pf_right, x)228 def findIntersection(fun0, fun1, x0):229 # one could also solve formula for the intersection230 # analytically, but fsolve is more flexible231 return fsolve(lambda x: fun0(x) - fun1(x), x0)232 middle = self.left + (self.right - self.left)/2233 touchdown = findIntersection(f_left, f_right, middle)[0]234 return touchdown235 def _set_touchdown_leave_auto(self, touchdown):236 self.iattributes.set_value('touchdown', touchdown,237 leave_automatic=True)238 def _set_touchdown_from_calibration(self):239 self.touchdown = self._calibration.touchdown240 @property241 def touchdown(self):242 return self.iattributes.touchdown243 @touchdown.setter244 def touchdown(self, touchdown):245 self.iattributes.touchdown = touchdown246 @property247 def _calibration(self):248 # calibration, which is used to:249 # - set touchdown from calibration250 # - visualize calibration.touchdown in GTouchdown251 return self.view_apply.calibration252 @property253 def pf_left(self):254 # get left datapoint means according to fit borders for fitting255 left_means = self.left_means256 # polynomial 2nd order order for negative positionZ257 return np.polyfit(left_means[:, 0], left_means[:, 1], 2)258 @property259 def pf_right(self):260 # get right datapoint means according to fit borders for fitting261 right_means = self.right_means262 # polynomial 2nd order for positive positionZ263 return np.polyfit(right_means[:, 0], right_means[:, 1], 2)264 @property265 def left_means(self):266 return self.bin_means[self.left_means_idx]267 @property268 def right_means(self):269 return self.bin_means[self.right_means_idx]270 @property271 def left_means_idx(self):272 x_check = self.bin_means[:, 0] < self.left273 y_check = self.bin_means[:, 1] < self.left_upper274 return np.logical_and(x_check, y_check)275 @property276 def right_means_idx(self):277 x_check = self.bin_means[:, 0] > self.right278 y_check = self.bin_means[:, 1] < self.right_upper279 return np.logical_and(x_check, y_check)280 @property281 def bin_means(self):282 if not hasattr(self, '_v_bin_means'):283 return None284 else:285 return self._v_bin_means286 @bin_means.setter287 def bin_means(self, bin_means):...
td_regression_candidates.py
Source:td_regression_candidates.py
1"""2Evaluates the TD regression candidates for a given position regarding3the mean. A high or low touchdown total is not predictive for future4performance as these players will often regress towards the mean or5average the following year.6"""7import os8import matplotlib.pyplot as plt9import numpy as np10import pandas as pd11import seaborn as sns12import tqdm13import src.preprocessing.playbyplay.playbyplay as pbp14from src.preprocessing.statistics.statistics import Statistics15sns.set_style("whitegrid")16def fix_player_names(name):17 split = name.split()18 first_initial = split[0][0]19 last_name = split[1]20 return '.'.join([first_initial, last_name])21if __name__ == "__main__":22 # year23 year = 202124 # position and play combinations25 comb = [("QB", "pass"), ("QB", "rush"), ("RB", "rush"), ("RB", "rec"), ("WR", "rec"), ("TE", "rec")]26 for position, play in comb:27 # load historical play-by-play data28 if not os.path.exists(f"../preprocessed/play-by-play/pbp_1999to{year}.csv"):29 df_prob = pbp.concat_playbyplay_data((1999, year))30 df_prob = df_prob.loc[:, ["rush_attempt", "rush_touchdown", "pass_attempt", "pass_touchdown",31 "yardline_100", "two_point_attempt", "year"]]32 else:33 df_prob = pd.DataFrame()34 chunks = pd.read_csv(f"../preprocessed/play-by-play/pbp_1999to{year}.csv", iterator=True, low_memory=False,35 chunksize=10000)36 for chunk in tqdm.tqdm(chunks):37 chunk = chunk.loc[:, ["rush_attempt", "rush_touchdown", "pass_attempt", "pass_touchdown",38 "yardline_100", "two_point_attempt", "year"]]39 df_prob = pd.concat([df_prob, chunk])40 # select data from previous years41 df_prob = df_prob.loc[(df_prob["year"] < year)]42 # select pass or run attempts43 if play == "pass" or play == "rec":44 # passing/receiving: pass attempt that ended in a touchdown45 df_prob = df_prob.loc[:, ["pass_attempt", "pass_touchdown", "yardline_100", "two_point_attempt"]]46 df_prob = df_prob.rename({"pass_attempt": "attempt", "pass_touchdown": "touchdown"}, axis=1)47 elif play == "rush":48 # rushing: rush attempt that ended in a touchdown49 df_prob = df_prob.loc[:, ["rush_attempt", "rush_touchdown", "yardline_100", "two_point_attempt"]]50 df_prob = df_prob.rename({"rush_attempt": "attempt", "rush_touchdown": "touchdown"}, axis=1)51 # select only normal attempts52 df_prob = df_prob.loc[(df_prob["two_point_attempt"] == 0) & (df_prob["attempt"] == 1)]53 # find probability of scoring a touchdown depending on distance to endzone54 df_prob = df_prob.groupby("yardline_100")["touchdown"].value_counts(normalize=True)55 df_prob = pd.DataFrame({"probability_of_touchdown": df_prob.values}, index=df_prob.index).reset_index()56 # keep stats where touchdown was scored and drop column57 df_prob = df_prob.loc[(df_prob["touchdown"] == 1)]58 df_prob = df_prob.drop("touchdown", axis=1)59 # plot probability of scoring a touchdown60 df_prob.plot(x="yardline_100", y="probability_of_touchdown", title=f"Probability for {play}ing TD")61 plt.savefig(f"../reports/td_regression_candidates/touchdown_probability_{play}.png")62 # load play-by-play data for defined year63 if not os.path.exists(f"../raw/play-by-play/play_by_play_{year}.csv"):64 df_train = pbp.get_playbyplay_data(year)65 else:66 df_train = pd.DataFrame()67 chunks = pd.read_csv(f"../raw/play-by-play/play_by_play_{year}.csv", iterator=True, low_memory=False,68 chunksize=10000, index_col=0)69 for chunk in tqdm.tqdm(chunks):70 df_train = pd.concat([df_train, chunk])71 # get player, team and distance to endzone72 if play == "pass":73 df_train = df_train.loc[:, ["passer_player_name", "posteam", "yardline_100"]]74 elif play == "rec":75 df_train = df_train.loc[:, ["receiver_player_name", "posteam", "yardline_100"]]76 elif play == "rush":77 df_train = df_train.loc[:, ["rusher_player_name", "posteam", "yardline_100"]]78 df_train = df_train.dropna()79 # rename features80 df_train = df_train.rename(columns={df_train.columns[0]: "player", "posteam": "team"})81 # join probability with training82 data = df_train.merge(df_prob, how="left", on="yardline_100")83 # group by player and team and sum up expected touchdowns84 data = data.groupby(["player", "team"], as_index=False).agg({"probability_of_touchdown": np.sum}).rename(85 {"probability_of_touchdown": "Expected touchdowns"}, axis=1)86 data = data.sort_values(by="Expected touchdowns", ascending=False)87 # rank the expected touchdowns88 data["Expected touchdowns rank"] = data["Expected touchdowns"].rank(ascending=False)89 # load final stats90 df_actual = Statistics(position, year).get_accumulated_data()91 # get player, team and td92 if play == "pass":93 df_actual = df_actual.loc[:, ["player", "team", "passing_td"]]94 df_actual = df_actual.rename({"passing_td": "touchdowns"}, axis=1)95 elif play == "rec":96 df_actual = df_actual.loc[:, ["player", "team", "receiving_td"]]97 df_actual = df_actual.rename({"receiving_td": "touchdowns"}, axis=1)98 elif play == "rush":99 df_actual = df_actual.loc[:, ["player", "team", "rushing_td"]]100 df_actual = df_actual.rename({"rushing_td": "touchdowns"}, axis=1)101 # clean up player names102 df_actual["player"] = df_actual["player"].apply(fix_player_names)103 # summarize the year104 df_actual = df_actual.groupby(["player", "team"]).agg({"touchdowns": np.sum}).reset_index()105 # merge raw and drop position106 data = df_actual.merge(data, how="left", on=["player", "team"]).dropna()107 # rename features and calculate rank of actual touchdowns108 data = data.rename({"touchdowns": "Actual touchdowns"}, axis=1)109 data["Actual touchdowns rank"] = data["Actual touchdowns"].rank(ascending=False)110 # calculate regression candidate111 data["Regression candidate"] = data["Expected touchdowns"] - data["Actual touchdowns"]112 data["Regression rank candidate"] = data["Actual touchdowns rank"] - data["Expected touchdowns rank"]113 # make simple true/false column for regression candidate114 data["Positive regression candidate"] = data["Regression candidate"] > 0115 # explore the statistics116 fig, ax = plt.subplots(figsize=(12, 8))117 ax.set_title(f"Positive regression candidates for {play} {position} in season {year}")118 # plot expected vs actual touchdowns and incorporate regression candidate values119 sns.scatterplot(120 x="Expected touchdowns",121 y="Actual touchdowns",122 hue="Positive regression candidate",123 data=data,124 palette=['r', 'g']125 )126 # plot line denoting border to positive/negative regression candidate127 max_actual_td = int(data["Actual touchdowns"].max())128 max_expected_td = int(data["Expected touchdowns"].max())129 max_td = max(max_actual_td, max_expected_td)130 sns.lineplot(x=range(max_td), y=range(max_td))131 # show top five negative and positive regression candidates132 data_sorted = data.sort_values(by="Regression candidate", ascending=False).reset_index()133 for i, row in data_sorted.iterrows():134 if i <= 1 or i >= len(data_sorted) - 1 - 1:135 ax.text(136 x=row["Expected touchdowns"] + 0.1,137 y=row["Actual touchdowns"] + 0.05,138 s=row["player"]139 )140 # save141 data_sorted.to_csv(142 f"../reports/td_regression_candidates/td_regression_candidates_{play}_{position}_{year}.csv",143 index=False144 )145 plt.savefig(146 f"../reports/td_regression_candidates/regression_candidates_{play}_{position}_{year}.png"...
TDRegressionCandidates.py
Source:TDRegressionCandidates.py
1# To add a new cell, type '# %%'2# To add a new markdown cell, type '# %% [markdown]'3# %%4import os5import sys6src_dir = os.path.join(os.getcwd())7abs_path = os.path.abspath(os.path.join(src_dir, os.pardir, 'src'))8sys.path.append(abs_path)9from utils import GLOBAL, functions10# %%11import pandas as pd12import nflfastpy as nfl13from matplotlib import pyplot as plt14import numpy as np15import seaborn as sns16sns.set_style('darkgrid')17# %%18#Change start and end for range19pbp_df = pd.DataFrame()20year_start = 201521year_end = 202022# %%23#Get roster based on last season, change date to last seasons24roster = nfl.load_2020_roster_data()25# %%26#Get yearly pbp data for past 5 years WARNING: Takes a long time27for year in range(year_start, year_end):28 yearly_df = nfl.load_pbp_data(year)29 pbp_df = pd.concat([pbp_df, yearly_df])30# %%31#Filter rushing data32#rush_attempt = 1 if it was a rush_attempt, same with rush_touchdown, & two_point_attempt = 0 to filter out 2pt conversions33rushing_df = pbp_df[['rush_attempt', 'rush_touchdown', 'yardline_100', 'two_point_attempt']]34rushing_df = rushing_df.loc[35 (rushing_df['two_point_attempt'] == 0) & (rushing_df['rush_attempt'] == 1)36]37# %%38#Filter receiving data39#pass_attempt = 1 if it was a rush_attempt, same with pass_touchdown, & two_point_attempt = 0 to filter out 2pt conversions40receiving_df = pbp_df[['pass_attempt', 'pass_touchdown', 'yardline_100', 'two_point_attempt']]41receiving_df = receiving_df.loc[42 (receiving_df['two_point_attempt'] == 0) & (receiving_df['pass_attempt'] == 1)43]44# %%45#Here, we are grouping by the yardline from where the play began, and then using value counts to count the number of times a rushing play was a touchdown (either a 0 or a 1), we can set the argument normalize = True to be able to calculate the proportion of plays that were touchdowns, instead of the count.46rushing_df_probs = rushing_df.groupby('yardline_100')['rush_touchdown'].value_counts(normalize=True)47rushing_df_probs = pd.DataFrame({48 'probability_of_td': rushing_df_probs.values49}, index=rushing_df_probs.index).reset_index()50#Filter out prob of not a td51rushing_df_probs = rushing_df_probs.loc[rushing_df_probs['rush_touchdown'] == 1]52rushing_df_probs = rushing_df_probs.drop('rush_touchdown', axis=1)53# %%54receiving_df_probs = receiving_df.groupby('yardline_100')['pass_touchdown'].value_counts(normalize=True)55receiving_df_probs = pd.DataFrame({56 'probability_of_td': receiving_df_probs.values57}, index=receiving_df_probs.index).reset_index()58#Filter out prob of not a td59receiving_df_probs = receiving_df_probs.loc[receiving_df_probs['pass_touchdown'] == 1]60receiving_df_probs = receiving_df_probs.drop('pass_touchdown', axis=1)61# %%62#Get PBP from the last season which is year_end variable63last_season_pbp_df = nfl.load_pbp_data(year_end)64# %%65#Filter out positions from last years rosters66#RBs67rb_df = roster.loc[roster['position'] == 'RB']['gsis_id']68#WRs69wr_df = roster.loc[roster['position'] == 'WR']['gsis_id']70#TEs71te_df = roster.loc[roster['position'] == 'TE']['gsis_id']72# %% [markdown]73# #RB TD Regression74# %%75#Get rushing data from last season pbp76last_season_RB_rushing_df = last_season_pbp_df.loc[last_season_pbp_df['rush_attempt'] == 1, ['rusher_id', 'rusher_player_name', 'posteam', 'rush_attempt', 'rush_touchdown', 'yardline_100']]77#Filter out RBs78last_season_RB_rushing_df = last_season_RB_rushing_df.loc[last_season_RB_rushing_df['rusher_id'].isin(rb_df)]79#Merge Probability df with Rushing80last_season_RB_rushing_df = last_season_RB_rushing_df.merge(rushing_df_probs, how='left', on='yardline_100')81#Calculate the actual touchdowns rb scored by aggregating all instances where the rush_touchdown == 1 & calculate aggregate of all instances where the probability of scoring a touchdown from that area of the field change names to better reflect what they represent82last_season_RB_rushing_df = last_season_RB_rushing_df.groupby('rusher_id', as_index=False).agg({83 'rusher_player_name': 'first',84 'rush_touchdown': np.sum,85 'probability_of_td': np.sum86}).rename({87 'probability_of_td': 'expected_touchdowns',88 'rush_touchdown': 'actual_touchdowns'89}, axis=1)90#Determine if positive regression candidate by comparing actual scored touchdowns vs expected touchdowns91last_season_RB_rushing_df['positive_regression_candidate'] = last_season_RB_rushing_df['actual_touchdowns'] < last_season_RB_rushing_df['expected_touchdowns']92#Get delta between expected vs actual tds93last_season_RB_rushing_df['delta'] = last_season_RB_rushing_df.apply(94 lambda x: abs(x.expected_touchdowns - x.actual_touchdowns), axis=195)96last_season_RB_rushing_df = last_season_RB_rushing_df.sort_values(by='expected_touchdowns', ascending=False)97# %% [markdown]98# #WR TD Regression99# %%100#Get receiving data from last season pbp101last_season_WR_receiving_df = last_season_pbp_df.loc[last_season_pbp_df['pass_attempt'] == 1, ['receiver_id', 'receiver_player_name', 'posteam', 'pass_attempt', 'pass_touchdown', 'yardline_100']]102#Filter out WRs103last_season_WR_receiving_df = last_season_WR_receiving_df.loc[last_season_WR_receiving_df['receiver_id'].isin(wr_df)]104#Merge Probability df with receiving105last_season_WR_receiving_df = last_season_WR_receiving_df.merge(receiving_df_probs, how='left', on='yardline_100')106#Calculate the actual touchdowns rb scored by aggregating all instances where the rush_touchdown == 1 & calculate aggregate of all instances where the probability of scoring a touchdown from that area of the field change names to better reflect what they represent107last_season_WR_receiving_df = last_season_WR_receiving_df.groupby('receiver_id', as_index=False).agg({108 'receiver_player_name': 'first',109 'pass_touchdown': np.sum,110 'probability_of_td': np.sum111}).rename({112 'probability_of_td': 'expected_touchdowns',113 'pass_touchdown': 'actual_touchdowns'114}, axis=1)115#Determine if positive regression candidate by comparing actual scored touchdowns vs expected touchdowns116last_season_WR_receiving_df['positive_regression_candidate'] = last_season_WR_receiving_df['actual_touchdowns'] < last_season_WR_receiving_df['expected_touchdowns']117#Get delta between expected vs actual tds118last_season_WR_receiving_df['delta'] = last_season_WR_receiving_df.apply(119 lambda x: abs(x.expected_touchdowns - x.actual_touchdowns), axis=1120)121last_season_WR_receiving_df = last_season_WR_receiving_df.sort_values(by='expected_touchdowns', ascending=False)122# %% [markdown]123# #TE TD Regression124# %%125#Get receiving data from last season pbp126last_season_TE_receiving_df = last_season_pbp_df.loc[last_season_pbp_df['pass_attempt'] == 1, ['receiver_id', 'receiver_player_name', 'posteam', 'pass_attempt', 'pass_touchdown', 'yardline_100']]127#Filter out WRs128last_season_TE_receiving_df = last_season_TE_receiving_df.loc[last_season_TE_receiving_df['receiver_id'].isin(te_df)]129#Merge Probability df with receiving130last_season_TE_receiving_df = last_season_TE_receiving_df.merge(receiving_df_probs, how='left', on='yardline_100')131#Calculate the actual touchdowns rb scored by aggregating all instances where the rush_touchdown == 1 & calculate aggregate of all instances where the probability of scoring a touchdown from that area of the field change names to better reflect what they represent132last_season_TE_receiving_df = last_season_TE_receiving_df.groupby('receiver_id', as_index=False).agg({133 'receiver_player_name': 'first',134 'pass_touchdown': np.sum,135 'probability_of_td': np.sum136}).rename({137 'probability_of_td': 'expected_touchdowns',138 'pass_touchdown': 'actual_touchdowns'139}, axis=1)140#Determine if positive regression candidate by comparing actual scored touchdowns vs expected touchdowns141last_season_TE_receiving_df['positive_regression_candidate'] = last_season_TE_receiving_df['actual_touchdowns'] < last_season_TE_receiving_df['expected_touchdowns']142#Get delta between expected vs actual tds143last_season_TE_receiving_df['delta'] = last_season_TE_receiving_df.apply(144 lambda x: abs(x.expected_touchdowns - x.actual_touchdowns), axis=1145)146last_season_TE_receiving_df = last_season_TE_receiving_df.sort_values(by='expected_touchdowns', ascending=False)...
Automation Testing Tutorials
Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.
LambdaTest Learning Hubs:
- JUnit Tutorial
- TestNG Tutorial
- Webdriver Tutorial
- WebDriverIO Tutorial
- Protractor Tutorial
- Selenium 4 Tutorial
- Jenkins Tutorial
- NUnit Tutorial
- Jest Tutorial
- Playwright Tutorial
- Cypress Tutorial
- PyTest Tutorial
YouTube
You could also refer to video tutorials over LambdaTest YouTube channel to get step by step demonstration from industry experts.
Run fMBT automation tests on LambdaTest cloud grid
Perform automation testing on 3000+ real desktop and mobile devices online.
Try LambdaTest Now !!
Get 100 minutes of automation test minutes FREE!!
Was this article helpful?
