File size: 17,871 Bytes
cf62ef7
 
 
4ee078c
cf62ef7
 
 
 
9147f5f
cf62ef7
12c5ad7
cf62ef7
4ee078c
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
cf62ef7
4ee078c
cf62ef7
 
4ee078c
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9147f5f
 
cf62ef7
 
9147f5f
cf62ef7
24ffe1d
9147f5f
 
 
 
 
24ffe1d
 
9147f5f
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
 
 
 
 
cf62ef7
 
 
 
 
12c5ad7
cf62ef7
4ee078c
cf62ef7
 
12c5ad7
cf62ef7
 
4ee078c
cf62ef7
 
12c5ad7
4ee078c
24ffe1d
4ee078c
 
 
 
 
 
cf62ef7
4ee078c
 
 
 
cf62ef7
12c5ad7
cf62ef7
 
 
 
12c5ad7
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
24ffe1d
 
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
 
 
 
cf62ef7
 
4ee078c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
24ffe1d
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
 
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
cf62ef7
4ee078c
 
 
 
cf62ef7
 
4ee078c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cf62ef7
 
 
 
 
 
 
 
 
 
 
24ffe1d
cf62ef7
 
 
 
4ee078c
cf62ef7
 
 
4ee078c
 
cf62ef7
 
 
4ee078c
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
 
cf62ef7
8317e1e
 
4ee078c
 
cf62ef7
 
4ee078c
cf62ef7
 
 
 
 
 
 
12c5ad7
 
 
 
8317e1e
12c5ad7
cf62ef7
 
24ffe1d
cf62ef7
4ee078c
12c5ad7
4ee078c
cf62ef7
12c5ad7
4ee078c
 
 
 
cf62ef7
 
12c5ad7
cf62ef7
 
4ee078c
24ffe1d
12c5ad7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9147f5f
af07826
 
12c5ad7
 
 
 
 
 
8317e1e
12c5ad7
 
 
 
 
8317e1e
12c5ad7
 
 
 
 
 
 
 
 
 
 
 
8317e1e
12c5ad7
 
 
 
 
 
 
 
 
 
 
 
24ffe1d
 
 
 
 
12c5ad7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4ee078c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
12c5ad7
4ee078c
 
 
 
 
 
 
 
 
 
cf62ef7
4ee078c
cf62ef7
4ee078c
cf62ef7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
"""
Baseball statistics application with txtai and Streamlit.

Install txtai and streamlit (>= 1.23) to run:
  pip install txtai streamlit
"""

import datetime
import math
import os
import random

import altair as alt
import numpy as np
import pandas as pd
import streamlit as st

from txtai.embeddings import Embeddings


class Stats:
    """
    Base stats class. Contains methods for loading, indexing and searching baseball stats.
    """

    def __init__(self):
        """
        Creates a new Stats instance.
        """

        # Load columns
        self.columns = self.loadcolumns()

        # Load stats data
        self.stats = self.load()

        # Load names
        self.names = self.loadnames()

        # Build index
        self.vectors, self.data, self.embeddings = self.index()

    def loadcolumns(self):
        """
        Returns a list of data columns.

        Returns:
            list of columns
        """

        raise NotImplementedError

    def load(self):
        """
        Loads and returns raw stats.

        Returns:
            stats
        """

        raise NotImplementedError

    def metric(self):
        """
        Primary metric column.

        Returns:
            metric column name
        """

        raise NotImplementedError

    def vector(self, row):
        """
        Build a vector for input row.

        Args:
            row: input row

        Returns:
            row vector
        """

        raise NotImplementedError

    def loadnames(self):
        """
        Loads a name - player id dictionary.

        Returns:
            {player name: player id}
        """

        # Get unique names
        names = {}
        rows = self.stats.sort_values(by=self.metric(), ascending=False)[["nameFirst", "nameLast", "playerID"]].drop_duplicates().reset_index()
        for x, row in rows.iterrows():
            # Name key
            key = f"{row['nameFirst']} {row['nameLast']}"
            key += f" ({row['playerID']})" if key in names else ""

            if key not in names:
                # Scale scores of top n players
                exponent = 2 if ((len(rows) - x) / len(rows)) >= 0.95 else 1

                # score = num seasons ^ exponent
                score = math.pow(len(self.stats[self.stats["playerID"] == row["playerID"]]), exponent)

                # Save name key - values pair
                names[key] = (row["playerID"], score)

        return names

    def index(self):
        """
        Builds an embeddings index to stats data. Returns vectors, input data and embeddings index.

        Returns:
            vectors, data, embeddings
        """

        # Build data dictionary
        vectors = {f'{row["yearID"]}{row["playerID"]}': self.transform(row) for _, row in self.stats.iterrows()}
        data = {f'{row["yearID"]}{row["playerID"]}': dict(row) for _, row in self.stats.iterrows()}

        embeddings = Embeddings(
            {
                "transform": self.transform,
            }
        )

        embeddings.index((uid, vectors[uid], None) for uid in vectors)

        return vectors, data, embeddings

    def metrics(self, name):
        """
        Looks up a player's active years, best statistical year and key metrics.

        Args:
            name: player name

        Returns:
            active, best, metrics
        """

        if name in self.names:
            # Get player stats
            stats = self.stats[self.stats["playerID"] == self.names[name][0]]

            # Build key metrics
            metrics = stats[["yearID", self.metric()]]

            # Get best year, sort by primary metric
            best = int(stats.sort_values(by=self.metric(), ascending=False)["yearID"].iloc[0])

            # Get years active, best year, along with metric trends
            return metrics["yearID"].tolist(), best, metrics

        return range(1871, datetime.datetime.today().year), 1950, None

    def search(self, name=None, year=None, row=None, limit=10):
        """
        Runs an embeddings search. This method takes either a player-year or stats row as input.

        Args:
            name: player name to search
            year: year to search
            row: row of stats to search
            limit: max results to return

        Returns:
            list of results
        """

        if row:
            query = self.vector(row)
        else:
            # Lookup player key and build vector id
            name = self.names.get(name)
            query = f"{year}{name[0] if name else name}"
            query = self.vectors.get(query)

        results, ids = [], set()
        if query is not None:
            for uid, _ in self.embeddings.search(query, limit * 5):
                # Only add unique players
                if uid[4:] not in ids:
                    result = self.data[uid].copy()
                    result["link"] = f'https://www.baseball-reference.com/players/{result["nameLast"].lower()[0]}/{result["bbrefID"]}.shtml'
                    result["yearID"] = str(result["yearID"])
                    results.append(result)
                    ids.add(uid[4:])

                    if len(ids) >= limit:
                        break

        return results

    def transform(self, row):
        """
        Transforms a stats row into a vector.

        Args:
            row: stats row

        Returns:
            vector
        """

        if isinstance(row, np.ndarray):
            return row

        return np.array([0.0 if not row[x] or np.isnan(row[x]) else row[x] for x in self.columns])


class Batting(Stats):
    """
    Batting stats.
    """

    def loadcolumns(self):
        return [
            "birthMonth",
            "yearID",
            "age",
            "height",
            "weight",
            "G",
            "AB",
            "R",
            "H",
            "1B",
            "2B",
            "3B",
            "HR",
            "RBI",
            "SB",
            "CS",
            "BB",
            "SO",
            "IBB",
            "HBP",
            "SH",
            "SF",
            "GIDP",
            "POS",
            "AVG",
            "OBP",
            "TB",
            "SLG",
            "OPS",
            "OPS+",
        ]

    def load(self):
        # Retrieve raw data from GitHub
        players = pd.read_csv("https://raw.githubusercontent.com/chadwickbureau/baseballdatabank/master/core/People.csv")
        batting = pd.read_csv("https://raw.githubusercontent.com/chadwickbureau/baseballdatabank/master/core/Batting.csv")
        fielding = pd.read_csv("https://raw.githubusercontent.com/chadwickbureau/baseballdatabank/master/core/Fielding.csv")

        # Merge player data in
        batting = pd.merge(players, batting, how="inner", on=["playerID"])

        # Require player to have at least 350 plate appearances.
        batting = batting[((batting["AB"] + batting["BB"]) >= 350) & (batting["stint"] == 1)]

        # Derive primary player positions
        positions = self.positions(fielding)

        # Calculated columns
        batting["age"] = batting["yearID"] - batting["birthYear"]
        batting["POS"] = batting.apply(lambda row: self.position(positions, row), axis=1)
        batting["AVG"] = batting["H"] / batting["AB"]
        batting["OBP"] = (batting["H"] + batting["BB"]) / (batting["AB"] + batting["BB"])
        batting["1B"] = batting["H"] - batting["2B"] - batting["3B"] - batting["HR"]
        batting["TB"] = batting["1B"] + 2 * batting["2B"] + 3 * batting["3B"] + 4 * batting["HR"]
        batting["SLG"] = batting["TB"] / batting["AB"]
        batting["OPS"] = batting["OBP"] + batting["SLG"]
        batting["OPS+"] = 100 + (batting["OPS"] - batting["OPS"].mean()) * 100

        return batting

    def metric(self):
        return "OPS+"

    def vector(self, row):
        row["TB"] = row["1B"] + 2 * row["2B"] + 3 * row["3B"] + 4 * row["HR"]
        row["AVG"] = row["H"] / row["AB"]
        row["OBP"] = (row["H"] + row["BB"]) / (row["AB"] + row["BB"])
        row["SLG"] = row["TB"] / row["AB"]
        row["OPS"] = row["OBP"] + row["SLG"]
        row["OPS+"] = 100 + (row["OPS"] - self.stats["OPS"].mean()) * 100

        return self.transform(row)

    def positions(self, fielding):
        """
        Derives primary positions for players.

        Args:
            fielding: fielding data

        Returns:
            {player id: (position, number of games)}
        """

        positions = {}
        for _, row in fielding.iterrows():
            uid = f'{row["yearID"]}{row["playerID"]}'
            position = row["POS"] if row["POS"] else 0
            if position == "P":
                position = 1
            elif position == "C":
                position = 2
            elif position == "1B":
                position = 3
            elif position == "2B":
                position = 4
            elif position == "3B":
                position = 5
            elif position == "SS":
                position = 6
            elif position == "OF":
                position = 7

            # Save position if not set or player played more at this position
            if uid not in positions or positions[uid][1] < row["G"]:
                positions[uid] = (position, row["G"])

        return positions

    def position(self, positions, row):
        """
        Looks up primary position for player row.

        Arg:
            positions: all player positions
            row: player row

        Returns:
            primary player positions
        """

        uid = f'{row["yearID"]}{row["playerID"]}'
        return positions[uid][0] if uid in positions else 0


class Pitching(Stats):
    """
    Pitching stats.
    """

    def loadcolumns(self):
        return [
            "birthMonth",
            "yearID",
            "age",
            "height",
            "weight",
            "W",
            "L",
            "G",
            "GS",
            "CG",
            "SHO",
            "SV",
            "IPouts",
            "H",
            "ER",
            "HR",
            "BB",
            "SO",
            "BAOpp",
            "ERA",
            "IBB",
            "WP",
            "HBP",
            "BK",
            "BFP",
            "GF",
            "R",
            "SH",
            "SF",
            "GIDP",
            "WHIP",
            "WADJ",
        ]

    def load(self):
        # Retrieve raw data from GitHub
        players = pd.read_csv("https://raw.githubusercontent.com/chadwickbureau/baseballdatabank/master/core/People.csv")
        pitching = pd.read_csv("https://raw.githubusercontent.com/chadwickbureau/baseballdatabank/master/core/Pitching.csv")

        # Merge player data in
        pitching = pd.merge(players, pitching, how="inner", on=["playerID"])

        # Require player to have 20 appearances
        pitching = pitching[(pitching["G"] >= 20) & (pitching["stint"] == 1)]

        # Calculated columns
        pitching["age"] = pitching["yearID"] - pitching["birthYear"]
        pitching["WHIP"] = (pitching["BB"] + pitching["H"]) / (pitching["IPouts"] / 3)
        pitching["WADJ"] = (pitching["W"] + pitching["SV"]) / (pitching["ERA"] + pitching["WHIP"])

        return pitching

    def metric(self):
        return "WADJ"

    def vector(self, row):
        row["WHIP"] = (row["BB"] + row["H"]) / (row["IPouts"] / 3) if row["IPouts"] else None
        row["WADJ"] = (row["W"] + row["SV"]) / (row["ERA"] + row["WHIP"]) if row["ERA"] and row["WHIP"] else None

        return self.transform(row)


class Application:
    """
    Main application.
    """

    def __init__(self):
        """
        Creates a new application.
        """

        # Batting stats
        self.batting = Batting()

        # Pitching stats
        self.pitching = Pitching()

    def run(self):
        """
        Runs a Streamlit application.
        """

        st.title("⚾ Baseball Statistics")
        st.markdown(
            """
            This application finds the best matching historical players using vector search with [txtai](https://github.com/neuml/txtai).
            Raw data is from the [Baseball Databank](https://github.com/chadwickbureau/baseballdatabank) GitHub project. Read [this
            article](https://medium.com/neuml/explore-baseball-history-with-vector-search-5778d98d6846) for more details.
        """
        )

        self.player()

    def player(self):
        """
        Player tab.
        """

        st.markdown("Match by player-season. Each player search defaults to the best season sorted by OPS or Wins Adjusted.")

        # Get parameters
        params = self.params()

        # Category and stats
        category = self.category(params.get("category"), "category")
        stats = self.batting if category == "Batting" else self.pitching

        # Player name
        name = self.name(stats.names, params.get("name"))

        # Player metrics
        active, best, metrics = stats.metrics(name)

        # Player year
        year = self.year(active, params.get("year"), best)

        # Display metrics chart
        if len(active) > 1:
            self.chart(category, metrics)

        # Run search
        results = stats.search(name, year)

        # Display results
        self.table(results, ["nameFirst", "nameLast", "teamID"] + stats.columns[1:] + ["link"])
        
        # Save parameters
        st.experimental_set_query_params(category=category, name=name, year=year)

    def params(self):
        """
        Get application parameters. This method combines URL parameters with session parameters.

        Returns:
            parameters
        """

        # Get parameters
        params = st.experimental_get_query_params()
        params = {x: params[x][0] for x in params}

        # Sync parameters with session state
        if all(x in st.session_state for x in ["category", "name", "year"]):
            # Copy session year if category and name are unchanged
            params["year"] = str(st.session_state["year"]) if all(params.get(x) == st.session_state[x] for x in ["category", "name"]) else None

            # Copy category and name from session state
            params["category"] = st.session_state["category"]
            params["name"] = st.session_state["name"]

        return params

    def category(self, category, key):
        """
        Builds category input widget.

        Args:
            category: category parameter
            key: widget key

        Returns:
            category component
        """

        # List of stat categories
        categories = ["Batting", "Pitching"]

        # Get category parameter, default if not available or valid
        default = categories.index(category) if category and category in categories else 0

        # Radio box component
        return st.radio("Stat", categories, index=default, horizontal=True, key=key)

    def name(self, names, name):
        """
        Builds name input widget.

        Args:
            names: list of all allowable names

        Returns:
            name component
        """

        # Get name parameter, default to random weighted value if not valid
        name = name if name and name in names else random.choices(list(names.keys()), weights=[names[x][1] for x in names])[0]

        # Sort names for display
        names = sorted(names)

        # Select box component
        return st.selectbox("Name", names, names.index(name), key="name")

    def year(self, years, year, best):
        """
        Builds year input widget.

        Args:
            years: active years for a player
            year: year parameter
            best: default to best year if year is invalid

        Returns:
            year component
        """

        # Get year parameter, default if not available or valid
        year = int(year) if year and year.isdigit() and int(year) in years else best

        # Slider component
        return int(st.select_slider("Year", years, year, key="year") if len(years) > 1 else years[0])

    def chart(self, category, metrics):
        """
        Displays a metric chart.

        Args:
            category: Batting or Pitching
            metrics: player metrics to plot
        """

        # Key metric
        metric = self.batting.metric() if category == "Batting" else self.pitching.metric()

        # Cast year to string
        metrics["yearID"] = metrics["yearID"].astype(str)

        # Metric over years
        chart = (
            alt.Chart(metrics)
            .mark_line(interpolate="monotone", point=True, strokeWidth=2.5, opacity=0.75)
            .encode(x=alt.X("yearID", title=""), y=alt.Y(metric, scale=alt.Scale(zero=False)))
        )

        # Create metric median rule line
        rule = alt.Chart(metrics).mark_rule(color="gray", strokeDash=[3, 5], opacity=0.5).encode(y=f"median({metric})")

        # Layered chart configuration
        chart = (chart + rule).encode(y=alt.Y(title=metric)).properties(height=200).configure_axis(grid=False)

        # Draw chart
        st.altair_chart(chart + rule, theme="streamlit", use_container_width=True)

    def table(self, results, columns):
        """
        Displays a list of results as a table.

        Args:
            results: list of results
            columns: column names
        """

        if results:
            st.dataframe(pd.DataFrame(results)[columns])
        else:
            st.write("Player-Year not found")


@st.cache_resource(show_spinner=False)
def create():
    """
    Creates and caches a Streamlit application.

    Returns:
        Application
    """

    return Application()


if __name__ == "__main__":
    os.environ["TOKENIZERS_PARALLELISM"] = "false"

    # Create and run application
    app = create()
    app.run()