Physical Characteristics of a MLB Player vs Their Performance
Data Management Final Performance Task
Table of Contents:
Introduction/Background 3
Methodology 4
Results 8
Future Studies 9
Conclusion 10
Bibliography 11
Appendices 12
Introduction/Background:
This report will focus on all the MLB groups and physical qualities of their players on the 2018 list. With association towards a group’s prosperity amid the season so far and a correlation towards past seasons previously. We will dissect the diverse attributes of stature, weight and age, just as estimating the results of accomplishment by observing the standings and win rates. We have made research inquiries to additionally look for the comprehension of this subject by inquiring:
- Does the average weight of the team affect their standings during a regular season
- What correlation is presented between the net points gained during the season and average weight of the team
- Does the relationship between the winning percentages correlate with the average weight of the top four standing teams from 20 years ago?
- Would a team be affected with their win percentage from the average age?
The reason behind this examination is to explore the shrouded biomechanics of players towards the game of baseball and on the off chance that it has an immediate impact towards their general execution. To answer these examination questions, We will look at the normal load of players on each group and their general standings for a season. Next, we will interface the average height (cm) to their number of runs scored amid a season and taking a gander at the association between the two. Alongside, going into profundity with win rates and the impacts from both age and weight, just as the connection between the insights of both the 2018 season and 2008 season. The target for all the examination information is, having the capacity to dissect the quality of the relationship and increase additional verification on our subject being either right, or off base.
A few people inside the overall population are devotees of baseball, be that as it may, don’t discuss the possibility of the impacts on physiological focal points for the game, which raised our interest. We trust that reliant on the game itself, there are sure focal points with regards to physiological improvement when playing for a long measure of time for one game, versus the benefit of strategy with another. Being that, there is no immediate association towards the physical characteristics a player may have, and overall performance, thus, we can put affirmation towards different variables that may be in favor of the game.
Get Help With Your Essay
If you need assistance with writing your essay, our professional essay writing service is here to help!
Bill Petti, from SB country made a comparable report, with motivation from a previous MLB proprietor and group president, Bill Veeck by seeing that MLB players kept up an explicit stature towards the most recent decade. Petti’s perceptions made the finish of vanquishing different ideas for different games which is, “greater is better” and permits considering different factors inside the game that assist with execution. Furthermore, additionally underscores that the game of baseball isn’t exclusively dependent on the physical make-up of the player. This exploration was open enough to help relate upon our point of physical characteristics, with expansion to execution and having the capacity to think about if his perceptions are right.
Methodology:
To get information for this examination, two dependable sources are found. The majority of the data were effectively available inside the assortment of sites found online that post about NBL measurements. Most of the data accumulated originated from the most dependable source which is the ESPN sports site and got information for normal weight, stature, age and standings through the program of each MLB group and their players for the 2018 season. Also, the data got from two sites called “group rankings” and “baseball reference” enabled me to experience win rates and winning rate for and age for the 2008 season. In the wake of gathering the information required for every factor inside our examination questions, We sorted out the data onto a spreadsheet to make tables and diagrams to then continue onto computing the relationship coefficient (shown in Figure 1) and slant.
Figure 1
Through examining the data, that were accumulated from the sources referenced above, we could concentrate on the connection between each physical characteristic and impact on execution. In the wake of watching the relationship on each diagram and calculating it, we could then decide the estimations of either positive or negative frail (0 to ± 0.33), moderate (± 0.33 to ± 0.66) and solid (±0.66 to 1). Dependent on these connections, We could then decide the quality of the connection between the two factors for every circumstance/question, which in the end displayed a method for disseminating the impact the factors had among one another in a strong or non steady path for our point. The utilization of visual portrayals of the information through the charts and tables themselves took into account to a greater extent a spread out rendition of the data that is expected to demonstrate our examination and appearing of a pattern, as assume to simply quantities of a connection. We have chosen to utilize the connection coefficient and incline (line of best fit) to show the connections for this point, as the data got together best suits this bit of information and theme of decision.
We made a rundown of inquiries that were connected upon our theme and requested that 15 irregular youths finish our overview. In the wake of doing as such, there were a reasonable sign that most of the outcomes originated from young people who are not really aficionados of baseball or play the game. The inquiries that were recorded, associated specifically to the exploration inquiries for our theme, which assisted with further understanding for open sentiments about our examination. The overview questions were:
- How often do you watch/follow baseball?
- Regularly
- Sometimes
- Never
- What type of physical characteristic do you think affects performance in the sport of baseball?
- Height
- Weight
- Age
- Other
- What age group do you expect to increase the likelihood of winning for the team?
- 25 and under
- 25 – 30
- 30 – 35
- 35 and older
- Based on the sport of baseball, what team do you think will be the most successful? The team with the best…
- Physical characteristics
- Best strategic skills
- Best team compatibility
- Coaching staff
- Do you think that the connection between the average height of a baseball player and winning percentage changed overtime?
- 25 and younger
- 25-30
- 30 +
(survey)^
Question: |
Answer #1 (%) |
Answer #2 (%) |
Answer # 3 (%) |
Answer # 4 (%) |
How often do you watch baseball? |
Regularly 10 |
Sometimes 0 |
Never 90 |
Only for the world series 0 |
What type of physical characteristic do you think affects performance in the sport of baseball? |
Height 10 |
Weight 10 |
Age 40 |
Other 40 |
What age group do you expect an increase for the likelihood of success |
25 and under 40 |
25-30 60 |
30+ 0 |
|
Do you think that the connection between the average height of a baseball player and winning percentage changed overtime?
|
Yes |
no |
Not sure |
|
Based on the sport of baseball, what team do you think will be the most successful? The team with the best…
|
Physical characteristics ZERO |
Strategic skills 50 |
Team collaboration 30 |
Coaching staff 20 |
The significance from these outcomes gave the proof on reinforcing our theory of this subject and thought of their comprehension towards the accomplishment of a group being for a real reason (advocated) and not the mindset of “greater is better”. A strategy that We utilized for this would be an arbitrary examining method, as there were a restriction on the measure of individuals that We needed to partake versus how much that really did influenced the outcomes. Because of me soliciting just a set number from individuals, caused not including contribution from others that may have a more grounded assessment about the point, which thus is comfort inspecting. Related upon this, the nearness of 2 predispositions were recognized. The overview contained an inspecting predisposition, on the grounds that the aggregate sum of individuals that partook were extremely constrained among the populace asked, just as were indicated towards an explicit age gathering and individuals who were in the center about the game. That, as well as the impediments with noting the inquiries also, made the decline of exactness for an individual answer with no genuine composed sentiment. Since, the appropriate responses were given to them to reply in a different decision organize, this caused a decrease in exactness for the outcomes. In conclusion, non-reaction predisposition were additionally experienced inside the review, since a few understudies did not consider the study important or chose to give off base reactions. Those negative ramifications can likewise have an immediate connection with the individuals who were given false data by others which can affect their reaction to the review.
Find Out How UKEssays.com Can Help You!
Our academic experts are ready and waiting to assist with any writing project you may have. From simple essay plans, through to full dissertations, you can guarantee we have a service perfectly matched to your needs.
View our academic writing services
The larger part of those that were overviewed on did not concur with the possibility of physical traits influencing execution. One explicit outcome on the best group to be effective with the best. Physical qualities were scored as zero, which were not what we had foreseen, being that the review depends on physical traits. In any case, being that there are different elements that add to the probability of accomplishment in execution for the help, the examination of this study is brought about physical attributes not influencing the achievement of the players on a group and backings our theory.
Results:
Subsequent to directing our examination, the outcomes were not amazing by any means, as it upheld our theory. Through foundation learning and comprehension of being a competitor, just as proof demonstrated here through these outcomes, indicate next to no connection between each physical trait and achievement/execution. The perceptions appeared through the diagrams and tables, alongside the counts of the connection coefficient and slant (appeared on page 12) give a viable aftereffect of no physical attributes being the purpose behind their execution amid the season. The insights even from the 2008 season contrasted with today make a similar consequence of no extreme result of progress. The main distinction between the two would be their age by and large. On the primary diagram (appeared on p. 12 figure 1.1-1.2) the normal weight (lbs.) versus standings made the normal result to be moderate solid, anyway made no connection at all between the two factors, making the relationship coefficient be (r = 0). This influences the aftereffect of this to wind up the inverse, which is as the load expands, the standings continue as before and have no effect on them. Additionally to this, the keep going chart on (page 15 figure 4.1-4.2) the examination between the normal age and win level of insights from 2008 season and 2018 season have no relationship between one another, causing a blunder inside the counts. Not exclusively is it in light of the fact that the two diagrams are viewed as discrete and are dissipated, however is a sort of proof, in view of the information gathered, that there is no relationship between the two factors. On the second diagram (appeared on page 13 figure 2.1-2.2) this displays a negative relationship between normal tallness (cm) and number of runs scored. Appearing solid negative relationship with (r = – 0.0000000000286), just as the third diagram (appeared on page 14 figure 3.1-3.2), however the inverse, being that it has a moderate positive connection of ( r = 0.512) between the factors, of normal tallness and win rate. This in any case, is exceptionally insignificant to be viewed as precise for the outcomes to be illustrative of the full relationship for progress, however bolsters the prior theory made.
A future expectation We would make for this point are that the outcomes will remain generally the equivalent, be that as it may, will increase some enhancement concerning the possibility of this subject. As insights are beginning to as of late give data of players programs to the MLB, it will push keep on ideally start some sort of discussion about physical credits and change in accordance with those new details for the game and a few favorable circumstances for it. In light of simply the age correlation as one of our examination question, it helps alot with how there is some change and the way that this game has been around for a long time, it is simply as of late that they are beginning to play around with measurements for the game and preferences all in all, which in our expectation will begin to be somewhat more updates in 20 years to come. That, as well as associations has changed throughout the years too, so demonstrates that the game is as yet enhancing after numerous years and may not be actually physical characteristics that the game will enhance, be that as it may, substantially more.
Future Studies:
All through this examination, there were numerous superfluous factors that could have affected the outcomes for the investigation of this examination. One would be that the factors itself were exceptionally restricted after attempting to look into the theme. Rather than having the free factor be the physical attributes of players on a group and their execution, different elements could ruin on their execution. To pick up a superior comprehension for explicit reasons with respect to why certain outcomes happened for various ways, different factors must be considered for to legitimize the reasons. Unessential factors may include:
- Coaching staff
- Positionings for type of player (size)
- Players experience (background)
- Technique
- Amount of training they do a week
Every one of the factors recorded, could have been a potential reason for why results inside our free and ward factors did not pursue an explicit pattern that were normal, This can help legitimize exceptions inside a diagram also which can be exhibited on (p… .) It is expected that the anomalies are caused by superfluous factors. On the off chance that there were an ideal pattern in the information, the presumption dependent on the study results and foundation learning about the game would have been the group with the most reduced estimation of an explicit trait would be positioned the most astounding in standings. Similarly, with the present exception in the information appeared on figure 1.1, exhibits that the foreseen finish of not being an impact can likewise be managed different components of the game. The comprehension of these anomalies gives the data as to some thinking behind why there is by all accounts almost no connection between every one of the factors, rather than impediments of mine.
On the off chance that this examination were to be rehashed, we would consider looking more into the group’s experience or the amount of a degree physical characteristics have for the game of baseball. By doing this, it would put incidental factors more into a controlled arrangement of information and doing additionally inquire about on those. That, as well as enhancing the correlation of the game as the years progressed. As referenced before, baseball has been a game that has been around for a long time, which contain measurements that enable us to relate present information to help improve a comprehension and precision of forecasts later on. The enhancement of this examination for taking a gander at the degree to which these factors influence each other will help for more precision on the objective of this examination.
Conclusion:
By experiencing factual examination inside this subject of physical traits versus execution in the MLB, it is resolved that the two factors don’t have a present connection. This is dependent on the counts made through finding the connection coefficient and demonstrating the pattern of every attribute, for example, tallness, weight and age, alongside win rate, runs scored and standings (achievement). This proof gave practically no connection between the factors (or the contrary instance of Figure 1.1) demonstrating the potential outcomes of unessential factors present causing an adjustment in one of the factors that would not have quite a bit of an impact between one another. The simple little esteemed connection coefficients on the chart gave a sign that the free factor of this examination (physical qualities) would not have enough capacity to complete a change for the outcomes in our exploration. It is apparent that our interest for this has achieved all the more fascinating pathways that interface with this point, nonetheless, is exceptionally upheld with our expectation of the impacts of physical qualities and relationship on execution for the game of baseball.
Works Cited
- “2008 Major League Baseball Season Summary.” Baseball-Reference.com, 26 July 2008, www.baseball-reference.com/leagues/MLB/2008.shtml.
- “2018 MLB Regular Season Standings Major League Baseball.” ESPN, ESPN Internet Ventures, www.espn.com/mlb/standings/_/group/overall.
- “Inside Pitch Magazine.” Official Magazine of the ABCA | Quarterly Publication for Baseball Players, Parents, Coaches & Fans, Inside Pitch, insidepitchonline.com/size-matters-not-in-baseball/.
- “Major League Rosters.” ESPN, ESPN Internet Ventures, www.espn.com/mlb/stats/rosters/_/sort/average_weight/order/true.
- “MLB Team Win % – All Games.” NBA Stats – NBA Team Points per Game on TeamRankings.com, 30 Mar. 2017, www.teamrankings.com/mlb/stat/win-pct-all-games.
Appendices:
Figure 1.1 Average weight (lbs) vs. standings (season result)
Figure 1.2
r = 0 → domain error
Figure 2.1 Average height (inches) vs. number of homeruns (runs scored)
r = -0.0000000000286 correlation
Figure 2.2
Figure 3.1 Average weight (lbs.) vs. win percentage (%)
Figure 3.1
r = 0.512 correlation
Figure 4.1 Relationship Between Winning Percentage & Average Age of 2018 vs. 2008
r = 0 → domain error
← 2008 not 1998
2008
2018
Cite This Work
To export a reference to this article please select a referencing style below: