How I Did It: John Vechey, Founder of PopCap

John Vechey 和他的伙伴们从辍学开始制作电脑游戏,到创立 PopCap 游戏公司并取得巨大成功的故事。他们推出的首个游戏 Bejeweled 销售超过五千万份,并创造了简单易玩的游戏理念。

It's a geek's dream come true. John Vechey dropped out of college in 1997 to work on a computer game with his pal Brian Fiete. This year, their Seattle-based game company, PopCap, which they founded in 2000 with Jason Kapalka, is on track to bring in $100 million. PopCap employs more than 300 and has offices in Shanghai, Seoul, and Dublin, with studios in San Francisco; Chicago; and Vancouver, British Columbia. The company's first of 35 games, Bejeweled, has sold more than 50 million units. PopCap's secret: Build games so accessible that anyone can play.

I grew up in Wisconsin. My dad's family worked in steel, but my parents were hippies. My dad taught me that it's not worth doing things you don't want to do -- he never worked more than a part-time job. For him, spending time with his friends, family, and women was more important than making money. He had priorities.

My parents got divorced when I was young. I lived with my mom and then moved to Indiana to live with my dad. I've had seven stepdads. My two half-siblings, though, are on my dad's side from his second marriage. He divorced again, and moved into my building in Seattle with my 16-year-old half-brother, who is at my place playing Xbox all the time. He plays all of our games before they're released.

I grew up pretty poor. I bond with anyone who has ever eaten government cheese. I never did extracurricular activities, because I always had a job. When I went to Purdue University, I met Brian Fiete in a programming class. I never had a computer growing up, but I wanted to be a computer engineer. Brian was always the first person done in class, and I was always second. I suggested we make a game together. It evolved into an online game we named ARC, based on an arcade version of paintball. At the time, my GPA was 1.67. I had to choose between working on this game or failing out of college. So I put all my energy into the game.

That was in 1997. People started playing our game online, and then someone named Warpig logged on and said, "Let's chat." That was Jason Kapalka, our third co-founder. He worked at a game company and wanted to license ARC.

We made $45,000, which, when you're 19 and from Indiana, feels like a million bucks. Meanwhile, some friends of the family were next-door neighbors with the founders of Sierra, a gaming company in Seattle. Someone at Sierra called us for an interview, and next thing we knew, Brian and I both left Indiana to work on games for Sierra. After we sold ARC to Sierra for $100,000, we left the company and used the money to start our own business with Jason.

It was not well thought out; more like, Let's live off ramen noodles, play games, and see what happens. We called our company Sexy Action Cool because we thought it was funny; plus, we thought we were going to develop games, not sell directly to the public. Jason and I were working on an animated PG-13 strip-poker game called Foxy Poker that had no nudity but was a really good game. We approached Strip-Poker.com, a porn site, and said, "Why don't you give us a bunch of money and sell our game?" They laughed and said no.

Then we created Bejeweled. I was in Indiana visiting family when I saw this simple solitaire game online -- no animation or graphics, but I thought it was cool. So I sent an e-mail to Brian and Jason with an idea for a game, which Brian created the next day using different colored circles. Jason sent a bunch of gem graphics on Day Three, and by Day Four, Bejeweled -- a really simple game where you match gems -- was done.

We tried to sell it to Pogo, the online gaming site. Yahoo didn't want it, either. We wound up making a flat-rate deal with Microsoft. It became phenomenally successful for MSN, with 60,000 users a day. But we were making only $1,500 a month.

Back then, in 2000, fans started asking for a downloadable version, because everyone was still using dial-up modems and didn't want to tie up their phone lines. So we made one, with better graphics and sound -- and charged for it. I had to convince Yahoo, MSN, and so on that people would play the free version on their sites and then download a better version for $20. And then we'd split the sale 50-50 with the host site. It was a new business model.

We launched in 2001 and made $35,000 the first month. The next month, we made $40,000. We were like, Holy crap! We're finally making money, but it won't last. So Brian and I hang out in Argentina and drink wine for four months. When Yahoo signed on, we moved back.

We didn't know anything about business, so we hired consultants who said, "We'll fix all your problems -- just pay us $100,000 and give us 3 percent of your company." That pissed us off -- if you don't play games, don't give advice on how to make games. They did get us to hire a comptroller. Before that, my aunt was doing the bookkeeping.

We decided from the start to make our games incredibly fun and easy so that they appeal to everyone. We currently have 35. We never think we have the magic formula or assume a hit. And yet every game we've done has made money. It took us three years to perfect Plants Vs. Zombies. We don't track the resources that go into each game. If it's a great game, it's worth a lot of money. If it's a B+ game, it's essentially worth zero.

In 2004, we had 15 employees and turned down a $60 million offer to buy our company. We knew we had to start taking the business side more seriously. So we hired David Roberts, our CEO. He had worked for Apple and Adobe, and he understood we wanted him to grow the business but leave the creative side alone.

When Dave started, we were focused on creating new games instead of supporting revenue streams from each game. Dave changed that. More than 30 percent of our annual revenue comes from Bejeweled, because it can be played on all these different platforms: PC/Mac, Xbox, PlayStation, Wii, DS, PalmPilot, iPhone, iPad, in-flight entertainment.

Jason and I interact a lot, and Jason and Brian interact a lot. Brian's working on a game with Jason, and Jason is also involved with the creative direction of the company. There's a power to three -- there's a constant rejiggering and shifting of opinions.

Facebook didn't even exist when we started the company. The iPhone didn't exist. We've adapted and changed and rolled with everything that's come our way. We've constantly integrated our approach to games. We're never perfect. We're always pretty good, but we're always trying to be better.

标题SpringBoot智能在线预约挂号系统研究AI更换标题第1章引言介绍智能在线预约挂号系统的研究背景、意义、国内外研究现状及论文创新点。1.1研究背景与意义阐述智能在线预约挂号系统对提升医疗服务效率的重要性。1.2国内外研究现状分析国内外智能在线预约挂号系统的研究与应用情况。1.3研究方法及创新点概述本文采用的技术路线、研究方法及主要创新点。第2章相关理论总结智能在线预约挂号系统相关理论,包括系统架构、开发技术等。2.1系统架构设计理论介绍系统架构设计的基本原则和常用方法。2.2SpringBoot开发框架理论阐述SpringBoot框架的特点、优势及其在系统开发中的应用。2.3数据库设计与管理理论介绍数据库设计原则、数据模型及数据库管理系统。2.4网络安全与数据保护理论讨论网络安全威胁、数据保护技术及其在系统中的应用。第3章SpringBoot智能在线预约挂号系统设计详细介绍系统的设计方案,包括功能模块划分、数据库设计等。3.1系统功能模块设计划分系统功能模块,如用户管理、挂号管理、医生排班等。3.2数据库设计与实现设计数据库表结构,确定字段类型、主键及外键关系。3.3用户界面设计设计用户友好的界面,提升用户体验。3.4系统安全设计阐述系统安全策略,包括用户认证、数据加密等。第4章系统实现与测试介绍系统的实现过程,包括编码、测试及优化等。4.1系统编码实现采用SpringBoot框架进行系统编码实现。4.2系统测试方法介绍系统测试的方法、步骤及测试用例设计。4.3系统性能测试与分析对系统进行性能测试,分析测试结果并提出优化建议。4.4系统优化与改进根据测试结果对系统进行优化和改进,提升系统性能。第5章研究结果呈现系统实现后的效果,包括功能实现、性能提升等。5.1系统功能实现效果展示系统各功能模块的实现效果,如挂号成功界面等。5.2系统性能提升效果对比优化前后的系统性能
在金融行业中,对信用风险的判断是核心环节之一,其结果对机构的信贷政策和风险控制策略有直接影响。本文将围绕如何借助机器学习方法,尤其是Sklearn工具包,建立用于判断信用状况的预测系统。文中将涵盖逻辑回归、支持向量机等常见方法,并通过实际操作流程进行说明。 一、机器学习基本概念 机器学习属于人工智能的子领域,其基本理念是通过数据自动学习规律,而非依赖人工设定规则。在信贷分析中,该技术可用于挖掘历史数据中的潜在规律,进而对未来的信用表现进行预测。 二、Sklearn工具包概述 Sklearn(Scikit-learn)是Python语言中广泛使用的机器学习模块,提供多种数据处理和建模功能。它简化了数据清洗、特征提取、模型构建、验证与优化等流程,是数据科学项目中的常用工具。 三、逻辑回归模型 逻辑回归是一种常用于分类任务的线性模型,特别适用于二类问题。在信用评估中,该模型可用于判断借款人是否可能违约。其通过逻辑函数将输出映射为0到1之间的概率值,从而表示违约的可能性。 四、支持向量机模型 支持向量机是一种用于监督学习的算法,适用于数据维度高、样本量小的情况。在信用分析中,该方法能够通过寻找最佳分割面,区分违约与非违约客户。通过选用不同核函数,可应对复杂的非线性关系,提升预测精度。 五、数据预处理步骤 在建模前,需对原始数据进行清理与转换,包括处理缺失值、识别异常点、标准化数值、筛选有效特征等。对于信用评分,常见的输入变量包括收入水平、负债比例、信用历史记录、职业稳定性等。预处理有助于减少噪声干扰,增强模型的适应性。 六、模型构建与验证 借助Sklearn,可以将数据集划分为训练集和测试集,并通过交叉验证调整参数以提升模型性能。常用评估指标包括准确率、召回率、F1值以及AUC-ROC曲线。在处理不平衡数据时,更应关注模型的召回率与特异性。 七、集成学习方法 为提升模型预测能力,可采用集成策略,如结合多个模型的预测结果。这有助于降低单一模型的偏差与方差,增强整体预测的稳定性与准确性。 综上,基于机器学习的信用评估系统可通过Sklearn中的多种算法,结合合理的数据处理与模型优化,实现对借款人信用状况的精准判断。在实际应用中,需持续调整模型以适应市场变化,保障预测结果的长期有效性。 资源来源于网络分享,仅用于学习交流使用,请勿用于商业,如有侵权请联系我删除!
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值